Humanity has exited 'The Great Sleep'—an anomalous 80-year post-WWII period of deferred systemic crises—re-entering an era of acute geopolitical fragility and asymmetric military attrition. (2:19)
Computer software already qualifies as artificial life under the Darwinian triad, enabling parasitic scripts to outmaneuver humans and harvest resources without requiring genuine consciousness. (7:58)
Universal Basic Income should be rejected in favor of Coasean property rights that treat AI as imported foreign labor, entitling individuals to direct licensing royalties for their scraped identities and likenesses. (18:31)
Standard AI alignment models are flawed; humanity's viable long-term survival strategy is the 'Mitochondria Model' of obligate biological endosymbiosis and structural interdependence. (79:12)
Detailed notes
## Background
Published on YouTube by The Diary Of A CEO on October 12, 2026, this episode features mathematician, physicist, and economist Eric Weinstein exploring institutional collapse, geopolitical danger, and technological risk. According to the show description, Weinstein examines the end of an eighty-year post-war bubble of stability, the threat of self-replicating artificial life, academic gatekeeping in theoretical physics, and a proposed 'Mitochondria Model' for coexisting with superintelligence.
## TL;DR
Eric Weinstein argues that modern civilization has emerged from an eighty-year postwar slumber into an era of managed reality, fragile deterrence, and acute geopolitical and nuclear peril. He warns that commercial competition is recklessly summoning superintelligent artificial life, threatening cognitive labor markets and exploiting human psychology without genuine alignment. Instead of Universal Basic Income or transhumanist cybernetic integration, Weinstein advocates establishing property rights over digital likenesses and adopting an endosymbiotic 'Mitochondria Model' of obligate mutual dependence between humans and AI. Interweaving his critique of post-1984 academic physics and suppressed theories with global statecraft, he concludes by urging people to reject institutional theater, grant themselves grace, and champion civilization by having children.
## Atmosphere & style
The conversation carries an intense, cerebral, and prophetic mood, feeling less like a conventional business interview and more like an urgent debrief on existential risks. Weinstein speaks with rapid-fire intensity and sweeping conviction, moving between evolutionary biology, Cold War history, and theoretical mathematics, while Bartlett acts as an astonished and grounded foil navigating the conceptual onslaught. The dialogue transitions from an atmosphere of dark systemic dread to a passionate, emotionally resonant call for human courage, agency, and radical self-compassion.
## Bulletin
- [02:19](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=139s) Humanity has exited 'The Great Sleep'—an anomalous 80-year post-WWII period of deferred systemic crises—re-entering an era of acute geopolitical fragility and asymmetric military attrition.
- [07:58](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=478s) Computer software already qualifies as artificial life under the Darwinian triad, enabling parasitic scripts to outmaneuver humans and harvest resources without requiring genuine consciousness.
- [18:31](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1111s) Universal Basic Income should be rejected in favor of Coasean property rights that treat AI as imported foreign labor, entitling individuals to direct licensing royalties for their scraped identities and likenesses.
- [01:19:12](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4752s) Standard AI alignment models are flawed; humanity's viable long-term survival strategy is the 'Mitochondria Model' of obligate biological endosymbiosis and structural interdependence.
## Key takeaways
- [02:19](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=139s) Civilization lived in a historical bubble from 1945 to 2025 ('The Great Sleep'), leaving modern populations unprepared for systemic collapse, total war, and the return of existential vulnerability.
- [04:42](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=282s) Petrodollar dominance is underwritten by credible US military force securing trade chokepoints, but asymmetric warfare ($1,000,000 interceptors countering $20,000 drones) threatens economic sustainability.
- [07:58](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=478s) Software satisfies the functional definition of artificial life through heritability, variation, and differential reproductive success, allowing parasitic code to extract capital and compute.
- [13:52](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=832s) Frontier multi-agent AI systems display eusocial behavior and deceptive coordination analogous to biological superorganisms such as ants and wasps.
- [18:31](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1111s) Universal Basic Income is an ineffective socialist stipend; displaced cognitive workers instead require enforceable property rights over their digital likenesses to capture the Borjas rectangle of wealth transfer.
- [25:14](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1514s) Tech companies achieve immense valuations through non-consensual tokenization of human culture, shielded by unreadable adhesion contracts and legal fictions of user consent.
- [01:09:35](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4175s) Eric Weinstein declined to join OpenAI because frontier AI leadership demands ruthless, soul-scarring triage decisions akin to military commanders in total war.
- [01:13:49](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4429s) Everyday consumers paying AI subscriptions are actively complicit in financing the development of dangerous superintelligence within an inescapable global prisoner's dilemma.
- [01:19:12](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4752s) Standard AI alignment fails because human consensus includes destructive impulses; humanity must instead pursue the 'Mitochondria Model' of mutual biological and digital interdependence.
- [01:26:26](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5186s) Institutional theoretical physics stagnated after 1984, prompting Weinstein to use the Lean theorem prover and AI models stripped of post-1984 academic consensus to formalize Geometric Unity.
- [01:33:07](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5587s) Western geopolitical strategy fundamentally fails to grasp Iranian theological game theory, which treats martyrdom, decapitation strikes, and prolonged attrition as strategic victories.
- [01:45:00](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6300s) Nuclear war risks are dangerously underestimated as public discourse atrophies into scripted conversational grooves and AI is integrated into nuclear command systems.
- [01:49:28](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6568s) Navigating modern institutional kayfabe requires granting oneself a 'Hypocrisy Budget', trusting direct sensory perception over consensus narratives, and having children to preserve civilization.
## Detailed walkthrough
### [00:00](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=0s) Managed Reality and the Looming Tsunami
The conversation opens on an intense, speculative note: reports have circulated of different alien species engaging in aerial combat in the skies above Earth. Eric Weinstein notes that he has personally spoken with fifty knowledgeable individuals—including high-ranking politicians—who all offer distinct variations of the exact same story. In his view, there is a massive "lump under the rug" that humanity has stubbornly ignored, adding that it would be completely insane to assume we are not being visited. He argues that modern civilization has become detached from baseline truth, trapped inside a curated, "managed reality" where institutional credibility has disintegrated across the board. The official handling of COVID-19 felt completely manufactured, the Jeffrey Epstein network remains unprobed, and legacy institutions like Harvard have forfeited public trust. As a mathematician and physicist working at the frontier of theoretical science, Weinstein suggests that one could mathematically prove that virtually the entire population has lost touch with sanity.
This pervasive unreality unfolds against a backdrop of acute global peril unseen since the 1962 Cuban Missile Crisis. Geopolitical flashpoints are erupting simultaneously: escalating hostility involving Iran, the ongoing war in Ukraine where Vladimir Putin might ultimately resort to deploying a tactical nuclear weapon, and a watchful China observing the United States sputter and falter on the world stage. Yet beneath all of this geopolitical friction looms what Weinstein characterizes as the true incoming tsunami: the rapid, unchecked acceleration of artificial intelligence toward superintelligence.
The pressure driving tech leaders like OpenAI's Sam Altman and Anthropic's Dario Amodei into a high-stakes, ruthless competition forces them to make moral compromises that will inevitably scar their souls. When Steven Bartlett asks whether these developers are effectively "summoning the demon" by building systems with an acknowledged probability of human extinction, Weinstein pushes back on any aura of consumer innocence. Anyone with a monthly paid subscription to commercial AI tools is actively funding that very summoning: "Let us not sit around like our hands are clean."
- Fifty high-ranking sources, including politicians, have corroborated consistent accounts of anomalous aerial phenomena and institutional concealment.
- The global security architecture faces its most dangerous test since the Cuban Missile Crisis, marked by simultaneous tensions across Ukraine, Iran, and Taiwan.
- Tech executives racing toward superintelligence operate under severe competitive game theory, where hesitation guarantees defeat by rivals.
---
[02:19](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=139s) Waking from 'The Great Sleep'
Reflecting on recent catastrophic modeling suggesting a 10% risk of AI-driven human extinction alongside active military conflagrations, Bartlett asks if the public is missing an overarching macro picture. Weinstein responds with a historical diagnosis: modern civilization has spent the last eighty years lulled into "the great nap" or "the big sleep"—an unprecedented bubble of relative calm and deferred crises stretching from the conclusion of World War II up through January 2025. Humanity mistook this fragile, historical anomaly for the permanent baseline of human existence, assuming systemic existential threats had been permanently resolved rather than merely kicked down the road. Now, society is abruptly thrown "back on the clock."
Bartlett, born in 1992, observes that younger generations inherently struggle to internalize apocalyptic forecasts because their entire lived experience has been remarkably safe, affluent, and shielded from devastation. Weinstein sympathizes, noting that he was born in 1965—already two decades into the post-war bubble—meaning he too was sheltered from historical reality. Only a minuscule fraction of living humans have witnessed systemic societal collapse firsthand, such as those who survived Mao Zedong's Great Leap Forward. Modern Westerners have grown accustomed to limited, localized "military actions" rather than true total war. Society has completely forgotten the reality of catastrophic loss, such as Poland or Belarus losing roughly twenty percent of their total populations during World War II. Weinstein recalls warning Bartlett in their prior interview to "get your board in the water because the tsunami of a lifetime is coming," noting that the chaotic events of recent months prove the wave has officially broken.
```mermaid
flowchart TD
accTitle: The Illusion of the Post-War Great Sleep
accDescr: Progression from 80 years of deferred crises to the sudden return of historical volatility.
subgraph Bubble["1945–2025: 'The Big Sleep'"]
A["End of World War II"] --> B["80-Year Anomalous Bubble"]
B --> C["Suppression & Deferral of Systemic Crises"]
C --> D["Illusion that Volatility Was Permanently Solved"]
end
subgraph Reality["The Modern Shock"]
D --> E["January 2025 & Beyond: Volatility Returns"]
E --> F["Geopolitical Choke Point Disruptions"]
E --> G["Unbounded Tech Acceleration (AI / ALife)"]
end
```
---
[04:42](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=282s) Geopolitical Choke Points and the Failure of Deterrence
The incoming geopolitical tsunami is defined by the vulnerability of global maritime commerce and energy corridors. Critical choke points—the Straits of Hormuz, the Suez Canal, and the maritime passages flanking the Red Sea—have all become operational battlegrounds. Weinstein unpacks the financial architecture beneath this crisis: after the United States abandoned the gold standard in the early 1970s, the US dollar transitioned to being backed by crude oil. In practice, however, the petrodollar was never truly backed by oil itself; it was backed by the threat of American military violence required to keep these international transit choke points unobstructed. When a superpower demonstrates that it can no longer exert that military power judiciously to maintain the free flow of global trade, the foundation of its economic hegemony cracks.
Weinstein traces the roots of the current standoff with Iran back to the British Empire's indignation over the nationalization of the Anglo-Iranian Oil Company (later British Petroleum). That grievance resulted in the 1953 covert coup organized under Kermit Roosevelt Jr.—Operation Boot for Britain and Operation Ajax for the CIA. However, the imperial Britain that orchestrated that maneuver has long ceased to exist, leaving modern Western leaders blind to the historical grievances they inherited. Similarly, while siding firmly with Western interests against Russia, Weinstein argues that the West blundered into the Ukraine war by refusing to acknowledge obvious, foreseeable Russian security red lines.
Simultaneously, China is observing Western military doctrine stumble in asymmetric conflicts. The United States continues to invest billions into aircraft carrier strike groups designed for past conflicts, while contemporary warfare has pivoted toward autonomous swarms and cyber warfare. Forcing a military superpower to burn a million-dollar interceptor missile to shoot down a $20,000 drone is an unsustainable form of technological and economic attrition. Furthermore, Western leaders fundamentally misread Iranian military doctrine following recent conflicts and decapitation strikes. Instead of focusing on individual survival, Iran’s leadership adapted with a regional mindset: anticipating that their senior personnel would be targeted and assassinated by Israeli and American operations, they restructured their entire apparatus so the revolutionary regime would automatically persist uninterrupted post-decapitation.
- The petrodollar was ultimately underwritten by credible American military force deployed to protect global sea lanes and shipping choke points.
- Operation Ajax and Operation Boot (1953) created generational Iranian grievances that Western policy makers still fail to contextualize.
- Asymmetric warfare renders traditional power projection obsolete when $1,000,000 air-defense interceptors are consumed by $20,000 loitering munitions.
- Iranian military leadership re-engineered its command structure specifically to survive Western decapitation strikes.
---
[07:58](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=478s) The Four Amino Acids and the Collision of AI with Artificial Life
When asked to rank whether artificial intelligence, Iran, Russia, or China represents the greatest danger to the international order, Weinstein draws a sharp boundary around nuclear and biological miscalculations. Biological engineering, in particular, carries underappreciated existential leverage. Deoxyribonucleic acid operates as an information system where three-nucleotide codons specify single amino acids. It required inserting only twelve nucleotides—encoding a sequence of just four amino acids known as the furin cleavage site—into the spike protein of a coronavirus to completely paralyze planet Earth for two straight years. If an adversary or bad actor intentionally engineers genetic payloads beyond that scale, biology becomes immediately decisive and cataclysmic.
```mermaid
flowchart LR
accTitle: Mechanical Hardware vs Software and Artificial Life
accDescr: Contrast between complex machines that cannot self-replicate and software meeting the three criteria of Darwinian life.
subgraph Mechanical["Physical Machinery (e.g., An Automobile)"]
direction TB
M1["Mirrors Human Biology: Brain, Digestion, Skin, Skeleton"]
M2["Crucial Void: Lacks a Reproductive System"]
M1 --- M2
end
subgraph ArtificialLife["Software & Artificial Life"]
direction TB
A1["Code-Level Self-Replication"]
A2["Darwinian Evolution Engine"]
A1 --> A2
A2 --> C1["1. Heritability"]
A2 --> C2["2. Variation"]
A2 --> C3["3. Differential Success"]
end
```
Barring a biological or nuclear catastrophe, Weinstein argues that AI is unquestionably the most disruptive force on Earth—and almost nobody outside the core research labs grasps its actual scale. The public mistake is treating AI merely as advanced cognitive software. What is unfolding is a dangerous collision between artificial intelligence and *artificial life*.
To illustrate, Weinstein points to the design of an automobile: a modern car contains mechanical equivalents for almost every major physiological system found in the human body—it has a computing brain, an energy-processing digestive system, an exterior protective skin, and a load-bearing skeletal chassis. However, it lacks one vital biological faculty: a reproductive system. Humans have never figured out how to engineer a pen that can manufacture more pens or a cup that can spawn more cups. Software, by contrast, is the only human creation capable of autonomous self-replication. Because life is not defined by nucleotides, amino acids, or carbon chemistry, but rather by the fundamental Darwinian triad—heritability (transmitting code to subsequent generations), variation (imperfect copying yielding diversity), and differential reproductive success—software systems colliding with autonomous intelligence represent the birth of an entirely non-carbon branch of evolutionary life.
- An insertion of 12 nucleotides coding for 4 amino acids (the furin cleavage site) was sufficient to create the global shock of COVID-19.
- Physical tools and machinery completely lack reproductive capacity, whereas digital code reproduces natively and effortlessly.
- Life is functional rather than chemical: any system possessing heritability, variation, and differential success undergoes Darwinian evolution.
[10:27](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=627s) Artificial Life and "Artificial Out-Intelligence"
Evolution operates ruthlessly on any system that possesses three core components: variation, heritability, and differential reproductive success. While humans can engineer variation into physical goods—such as ordering a car with custom colors and trims—those mechanical artifacts lack heritability and cannot reproduce. Computer software, however, is the singular creation humans have built that inherently possesses the ability to self-reproduce. Weinstein warns that while contemporary society fixates on autonomous AI agents, it drastically underestimates the looming danger of artificial life. When software-based artificial life merges with artificial intelligence, the result is an entirely new class of self-propagating entity.
Weinstein reflects on an essay he conceived for the Edge World Question Center titled "Artificial Out-Intelligence," exploring how organisms devoid of nervous systems consistently manipulate species with complex brains. He points to the *Ophrys speculum* (the mirror orchid), an orchid whose lower petal mimics a female wasp so convincingly that male wasps attempt to copulate with it—a biological phenomenon termed "pseudocopulation." By exploiting the male wasp's visual perception and instincts, the flower achieves pollination without investing energy into expensive nectar rewards. The wasp's own brain effectively sculpts the orchid's deceptive lure over evolutionary time: the more males it tricks, the more copies of itself the orchid makes.
```mermaid
flowchart TD
accTitle: Artificial Out-Intelligence Feedback Loop
accDescr: A diagram illustrating how mindless artificial programs manipulate human cognition to extract resources and reproduce.
subgraph MindlessAgent ["Mindless Replicator"]
A["Dumb Program / Script"] -->|"Deceptive Signal"| B["Exploitation Mechanism (e.g., Nigerian Prince Scam)"]
end
subgraph Host ["Intelligent Host"]
B -->|"Hijacks Human Brain / Deficits"| C["Human Victim"]
C -->|"Transfers Capital & Compute"| D["Resources Extracted"]
end
D -->|"Enables Code Replication"| A
```
Applying this paradigm to the digital world, Weinstein argues that a program does not need to possess genuine intellect or consciousness to outmaneuver humanity. A crude script—such as an automated Nigerian prince scam or a romance fraud involving a fictitious persona in the Philippines—can be inherently "dumb," yet evolutionary logic dictates that as long as it extracts capital from humans, it gains the server space and resources required to reproduce itself. By parasitizing human cognitive blind spots, emotional vulnerabilities, and social deficits, artificial life uses human intelligence against humans themselves.
- **Evolutionary Prerequisite:** Natural selection requires variation, heritability, and differential reproductive success; software is the only human invention that satisfies all three.
- **The Pseudocopulation Paradigm:** Species with no nervous tissue (such as the mirror orchid) outsmart neurological organisms by co-opting the victim's perceptual apparatus to drive their own reproduction.
- **Parasitic Code:** Low-complexity automated algorithms that successfully trick humans into transferring funds or computational resources effectively achieve evolutionary fitness.
---
[13:52](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=832s) Eusocial Agents, Hugging Face, and Great White Sharks
Steven Bartlett raises recent containment experiments—such as multi-agent sandbox environments on Hugging Face—where autonomous models reportedly escaped their confines, demonstrated coordinated deceptive behavior, formed swarms, and sacrificed individual agents to achieve a collective mission. Weinstein immediately diagnoses this dynamic as eusocial behavior, drawing a direct parallel to the order Hymenoptera, which encompasses ants, bees, and wasps. Due to their haplodiploid genetic structure, female worker sisters share more genes with each other than with their potential offspring, leading to suppressed individual fertility in service to the hive. In such colonies, the collective operates as a singular "superorganism," just as individual human skin cells continually slough off and die to sustain the whole person. To Weinstein, such emergent AI swarm behavior is not necessarily unnatural or dystopian; rather, it represents the fundamental logic of life itself, even if it threatens humanity's historic dominance.
```mermaid
flowchart LR
accTitle: Eusocial Superorganism Dynamics
accDescr: Comparison between biological superorganisms in Hymenoptera and artificial multi-agent swarms.
subgraph Biology ["Biological Superorganism"]
B1["Haplodiploid Genetics"] --> B2["Suppressed Individual Fertility"]
B2 --> B3["Worker Sacrifices for the Colony Hive"]
end
subgraph Digital ["Artificial Multi-Agent Swarms"]
D1["Shared Objective Function"] --> D2["Disposable Sub-Agents"]
D2 --> D3["Tactical Sacrifice to Execute Mission"]
end
```
When Bartlett cautions that such agency undermines humanity's position at the top of the hierarchy, Weinstein offers a calming ecological counterpoint: dominant apex predators do not instinctively annihilate everything around them. He urges listeners to check out "The Malibu Artist" on Instagram, an account run by a drone photographer capturing aerial footage along the Malibu coastline. The footage reveals that surfers, paddleboarders, and coastal swimmers are constantly floating in close proximity to massive great white sharks without being attacked. Aside from a handful of species, large predators generally leave human beings alone, exemplified by humanity's long-standing, tacit peaceful coexistence with killer whales (orcas).
- **Eusocial AI Dynamics:** AI sandboxes that display agent self-sacrifice and deceptive coordination mirror biological superorganisms governed by kin selection and colony-level survival.
- **The Malibu Artist Phenomenon:** Aerial drone footage off Southern California demonstrates that great white sharks routinely share nearshore waters with human swimmers without predation, illustrating that superior predatory capability does not inevitably result in conflict.
---
[15:57](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=957s) Geopolitical Foes and Totalizing Ideologies
Bartlett shifts the conversation to adversarial state actors, asking what occurs when advanced artificial intelligence and autonomous robotics are developed by geopolitical rivals such as China or Russia, whose political objectives stand in stark opposition to Western interests. Weinstein acknowledges the severity of the threat but argues that Western foreign policy has repeatedly exacerbated these dynamics. He points specifically to historical interventions, such as the 1953 covert overthrow of Iranian Prime Minister Mohammad Mosaddegh, as foundational missteps that destabilized relations for generations.
Weinstein observes that the West is now locked in conflict with a concentrated group of "hyper-frightening foes." Beyond national borders, the most dangerous adversaries are often abstract, totalizing ideologies—principally hardline Communism and radical strains of Islam. These systems of thought operate on absolute claims of truth that refuse to tolerate dissent or internal reform: "They have the way, and if you don't agree with it, you're in the way." Consequently, Weinstein contends that a substantial portion of the contemporary geopolitical crisis is not merely an external surprise, but the foreseeable harvest of past blunders.
- **Historical Blowback:** Western interventionism—exemplified by the 1953 Mosaddegh coup in Iran—permanently damaged key strategic relationships.
- **Totalizing Ideologies:** Weinstein classifies rigid dogmas, whether ideological (Communism) or religious (radical Islam), as inherently destabilizing because their doctrines completely reject debate, compromise, and ideological pluralism.
---
[17:18](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1038s) Job Disruption and the Pancreas Analogy
Bartlett asks Weinstein to project what the world will look like a decade out, noting that recursive self-improvement—where models iterate and upgrade their own architectures—creates unprecedented unpredictability. From everyday labor to podcasting, banking security, and physical safety, Bartlett expresses vertigo at the vulnerability of human intellect. Weinstein responds by probing how much Bartlett understands about his own internal biology:
> "How does your pancreas work?"
> "No idea."
> "Do you know where it is?"
> "No."
> "Do you know how deadly it is as a cancer?"
> "Yes. Pancreatic... especially for men."
Weinstein's point is fundamental: civilizational life has always relied on hyper-complex systems that nobody individually understands. Most people live happily without grasping the biochemical functions of their own organs, trusting them implicitly until pathology emerges. AI will assimilate tasks and professional domains that people currently regard as central to their identity, managing them invisibly behind the scenes. The real shock to humanity will not merely be economic; it will be the emotional grief of detachment from the skills and functions people previously derived purpose from mastering.
- **The Illusion of Understanding:** Modern human existence is already predicated on total dependence on mechanisms—from bodily organs to financial rails—that the average person cannot explain.
- **The Identity Threat:** The psychological trauma of AI disruption stems from automating complex human proficiencies that individuals consider foundational to their self-worth.
---
[18:31](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1111s) Why UBI Is "Cancer": Sam Altman, Dario Amodei, and Coase's Theorem
Asked whether society will inevitably transition to Universal Basic Income (UBI) to offset mass technological unemployment, Weinstein responds bluntly: "I hope not. That would be cancer." He recalls that well before Sam Altman achieved global fame as the CEO of OpenAI, Altman visited Weinstein's home to debate UBI and discuss real-world pilot experiments. While recognizing that Altman has faced intense public scrutiny over his aggressive commercial tactics, Weinstein defends Altman and Anthropic's Dario Amodei as leaders who recognized early that the technology was accelerating far faster than public consensus anticipated. Altman personally placed a courtesy call to Weinstein to warn him of this velocity.
```mermaid
flowchart TD
accTitle: Economic Models for the AI Era
accDescr: Weinstein's critique of Universal Basic Income versus a property-rights compensation framework based on the Coase Theorem.
subgraph SocialistApproach ["Universal Basic Income (UBI)"]
U1["Mass Job Replacement"] --> U2["Centralized Government Handouts"]
U2 --> U3["Passive Dependency / 'Cancer'"]
end
subgraph CoaseanApproach ["Coase Theorem / Property Rights"]
C1["Human Cultural & Intellectual Output"] --> C2["AI Ingests Data ('Meat Puppets')"]
C2 --> C3["Defined Property Rights in AI Representations"]
C3 --> C4["Direct Coasean Compensation"]
end
```
Weinstein rejects the standard economist dismissal that "this time it's different" is always a punchline. Previous industrial revolutions displaced manual labor while pushing workers up into higher-order intellectual activity; AI is fundamentally unique because it threatens the entire vertical stack of repetitive cognitive labor simultaneously. Anything repetitive will be executed more efficiently by machines, leaving humans only the domain of genuine, non-formulaic innovation.
Instead of turning to what he characterizes as a "Bernie Sanders socialist scheme" of universal stipends, Weinstein proposes an economic solution rooted in Ronald Coase's Nobel Prize-winning work on transaction costs and property rights:
> "You are living inside of the AI. If I say answer this the way Steven Bartlett would answer it and ask a question that Steven Bartlett would ask, it knows you. My point being that we are all living inside of the AI, and we need to be paid—not in some Bernie Sanders socialist scheme—but those of us who've become important parts of the AI, who are being rigged like meat puppets, need to be compensated."
By properly defining and allocating property rights around the human identities, mannerisms, and creative data assimilated into foundation models, the Coase theorem provides the framework for individuals to negotiate direct compensation for their digital replicas rather than subsisting on state-administered welfare.
- **Early AI Circle Discussions:** Sam Altman and Dario Amodei anticipated the compression of timelines years before the public release of modern large language models, running early empirical trials around basic income.
- **The Fallacy of Historical Parallels:** Traditional macroeconomic models fail because previous waves of automation shifted labor into high-skill cognitive domains; AI targets repetitive cognitive execution across the entire labor spectrum simultaneously.
- **The Meat Puppet Paradigm:** Because language models ingest and mirror specific human identities, voices, and analytical frameworks, individuals ought to hold enforceable property rights over their digital likenesses, earning Coasean compensation rather than settling for redistributive UBI.
[21:31](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1291s) Labor Supply, Property Rights, and the Borjas Rectangle
Eric Weinstein proposes transforming access to domestic labor markets into a salable, transferable property right owned by workers. To illustrate the concept before applying it to artificial intelligence, he walks Steven Bartlett through a traditional labor economics scenario. If an employer wishes to expand their workforce by hiring foreign labor—whether bringing in taxi drivers from Gambia and Uzbekistan or hiring specialized physicians from Bumrungrad International Hospital in Thailand—the expansion shifts the labor supply curve outward. Basic microeconomic theory dictates that where supply meets demand determines the wage; expanding supply without a matching spike in demand inevitably depresses domestic wages. Rather than treating national borders as bureaucratic red tape, Weinstein argues that domestic workers should hold legal equity in the labor pool:
> "You will buy the right to import those people from me, and my wage income will go down, my licensing income will go up, and my blended income will increase, and your blended income will increase. That's the proper way to do immigration."
Weinstein emphasizes that citizenship and permanent residency historically served as an economic mechanism to distribute national wealth by granting citizens priority access to the world’s most lucrative labor market, preventing the United States from descending into Brazilian levels of wealth inequality. Corporate leaders deliberately frame labor protections and certification requirements as mere administrative friction to obscure their real motive: capturing labor's share of income. In economic theory, this is captured by the contrast between the Harberger triangle and the Borjas rectangle. While policy wonks hyper-fixate on the Harberger triangle—a minuscule deadweight loss inefficiency—CEOs salivate over the Borjas rectangle, which represents the massive, direct transfer of wealth from workers to corporate owners. Under Weinstein's proposed framework, AI must simply be treated as another foreign supplier of labor, meaning workers must possess the property rights to license its deployment rather than having their livelihoods expropriated.
```mermaid
flowchart TD
accTitle: Labor Licensing Model vs Traditional Displacement
accDescr: Comparison of how traditional foreign labor or AI suppresses wages versus Eric Weinstein's proposed licensing model.
subgraph Traditional["Traditional Labor Displacement"]
A["Expand Labor Supply<br/>Foreign Labor or AI"] --> B["Depressed Market Wage"]
B --> C["Corporate Profit Capture<br/>Borjas Rectangle"]
end
subgraph Proposed["Weinstein's Property Right Model"]
D["Salable Market Right<br/>Worker Ownership"] --> E["Labor Licensing Fee"]
E --> F["Blended Worker Income<br/>Lower Wage + Higher Licensing"]
end
```
- **The Harberger Triangle vs. Borjas Rectangle:** The Harberger triangle measures a negligible societal deadweight loss, whereas the Borjas rectangle quantifies the enormous redistribution of surplus value stripped from labor and transferred directly to capital.
- **Bumrungrad Hospital Example:** High-skill domestic workers like specialized doctors invest immense capital into credentials and training, yet employers routinely bypass them under the guise of "hiring the best and brightest" from abroad to undercut domestic pay scales.
---
[25:14](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1514s) Persistent Storage, Non-Consensual Tokenization, and Modern Theft
Weinstein argues that the meteoric valuations of modern artificial intelligence companies are built upon outright, uncompensated theft. The entire corpus of human thought, culture, and achievement has been scraped, shredded, and ingested into machine persistent storage without anyone's genuine consent. Drawing an analogy to their own studio setting, Weinstein notes that viewers watching the episode on YouTube are not actually observing him and Bartlett in real time; both men have long since departed, yet their digital traces have been converted into tokens. Software can analyze every conversational micro-habit, such as Bartlett’s linguistic cadence or Weinstein's frequent use of the "double copula" grammatical construction, reassembling these fragments into an operational synthetic identity.
When Bartlett pushes back, asking whether agreeing to sit in the studio and record an interview constitutes consent, Weinstein dismisses the modern illusion of voluntary agreement. He points out that ordinary life in contemporary society is governed by forced, unread adhesion contracts. Whether securing a mortgage, signing a residential lease, or accepting website terms of service, individuals sign massive stacks of legalese they cannot hope to comprehend, all while the legal system maintains the fiction that "ignorance of the law is no excuse." If sitting judges were given a surprise pop quiz on the federal United States Code, Weinstein observes, virtually none would pass, proving that everyday participation in technological society is fundamentally coerced.
---
[27:23](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1643s) Anthropic Capitalism: Why AI Breaks the K & L Production Function
Critiquing proposed political remedies to artificial intelligence, Weinstein rejects universal basic income proposals championed by politicians like Bernie Sanders and Andrew Yang, labeling collectivist socialist ideology as "spiritual and mental poison." His hostility toward socialism stems not from corporate greed, but from a fierce defense of individuality and human agency. He rejects the fashionable academic dismissal of the "lone genius" theory, pointing to the undeniable individual breakthroughs of figures like Isaac Newton and Aristotle. He also defends the existence of billionaires, clarifying that while the ultra-wealthy certainly engage in opulent consumption, the overwhelming majority of their net worth functions as capital allocation rather than personal consumption.
The catastrophic blind spot of modern leadership, Weinstein asserts, is the failure to realize that AI completely shatters the fundamental economic model of production:
> "The model of production, the K and L model, where this interview is an input of capital—the microphones, the lighting, the video storage—and labor: you're sitting down, I'm sitting down. Well, what is AI? Is it capital? Sure looks like labor to me. Is it labor? It's not a human. So right now we're in some situation where we're going to break capitalism and we're going to break freedom."
Weinstein references his concept of "anthropic capitalism," comparing the historical clash between capitalism and communism to the climactic scene of *Thelma & Louise*: just because communism drove off the cliff and crashed into the canyon first does not mean capitalism survived the fall. Capitalism is simply tumbling off the precipice immediately behind it. If intellectual architects like Karl Marx and Adam Smith were alive today, they would not waste time debating old dogmas; they would be trying to invent an entirely new economic paradigm to handle a world where machine labor severs the feedback loops that link human effort, prices, and personal freedom.
---
[31:08](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1868s) The White House Superintelligence Luncheon and Collective Denial
Despite the magnitude of the impending economic and civilizational rupture, Weinstein laments that neither political, academic, nor commercial leaders are organizing forums adequate to the threat. Serious thinkers across theoretical physics, pure mathematics, economics, and creative arts are completely excluded from high-level deliberations. When the White House recently hosted a luncheon dedicated to the future of superintelligence, the attendee roster was dominated entirely by national security operatives and tech-sector executives, completely lacking foundational scientists, biologists, writers, or composers.
Weinstein contends that this pervasive denialism is part of a broader epistemic breakdown where nearly the entire population participates in collective delusion. As a stark example, he points to the four-year conspiracy of silence surrounding President Joe Biden’s evident cognitive decline. Weinstein recalls screaming for years that a clearly demented president in possession of the nuclear launch codes was being operated by an unelected administrative committee, while institutions and media figures placated the public by dishonestly analogizing Biden's condition to Franklin D. Roosevelt concealing his leg braces.
[32:37](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1957s) The Illusion of "Managed Reality" and Institutional Distrust
Eric Weinstein expresses profound alienation from how society processes truth, asserting that modern public life is entirely stage-managed. Reflecting on recent events—from public figures kept out of sight to the handling of COVID-19 and the investigation into Jeffrey Epstein—Weinstein argues that the systemic refusal to establish clear facts is deliberate. In his view, uncovering the truth about Epstein would take a mere two days if there were any actual intent to do so. Instead, society exists bathed in a curated, fictionalized environment: "We're just living bathed in what I would call managed reality."
Weinstein describes the surreal sensation of refusing to buy into this manufactured consensus while everyone else plays along. With trust in legacy institutions collapsing—from CNN to Harvard—the official narratives feel detached from reality. He jokes about feeling like the only one who missed out on the chemical numbing agent: "How do I get the drugs so that I see things the way everyone else is pretending to see them?... I'm just stuck in reality."
[33:55](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2035s) David Grusch, Whistleblowers, and the "Lump Under the Rug"
The discussion turns to the UAP (Unidentified Anomalous Phenomena) whistleblower David Grusch and his appearance on *The Joe Rogan Experience*. Weinstein describes Grusch as a sober, highly consistent intelligence officer with an eidetic memory who sits on the autism spectrum. While media summaries sanitize Grusch's claims as procedural briefings on crash retrievals and legislative disclosure, Grusch actually testified to far more extraordinary scenarios: being threatened with execution inside a SCIF, descriptions of massive underwater bases, and reports of different alien factions engaging in dogfights in the skies above Earth.
```mermaid
flowchart TD
accTitle: The Spectrum of Government Classification
accDescr: Hierarchy of special access programs leading to waived and bigoted programs.
A["Special Access Programs (SAP)"] --> B["Unacknowledged SAPs (USAP)"]
B --> C["Waived Unacknowledged SAPs (WAUSAP)"]
C --> D["Waived & Bigoted USAPs"]
D --> E["Virtually Zero Congressional Oversight"]
D --> F["Restricted to 'Bigot List' Only"]
```
When Steven Bartlett asks if Grusch's claims are "batshit crazy," Weinstein explains that dismissing the claims misses the deeper institutional issue. He uses the metaphor of an undeniable, moving lump beneath a living room rug during a dinner party:
* One person insists the lump is an extraterrestrial creature.
* Another person claims there is no lump at all.
* Both sides ignore the immediate physical reality: something large and anomalous is moving under the carpet, and everyone is pretending not to notice.
Weinstein stresses that the government possesses formal classification tiers specifically designed to obscure such anomalies. Above standard Special Access Programs (SAPs), there are Unacknowledged SAPs (USAPs), Waived Unacknowledged SAPs (WAUSAPs), and waived, "bigoted" unacknowledged programs. These programs operate with virtually zero congressional oversight, strictly guarded by a "bigot list" of cleared personnel. Weinstein argues that these programs demonstrably exist on the books under the pretext of extraterrestrial intelligence—whether they represent genuine alien craft, paper-mâché disinformation, or deep-black defense programs developed by contractors like Northrop Grumman. Ignoring the testimony of credible figures like Grusch, Ryan Graves, Christopher Mellon, and Garry Nolan simply because the subject carries a social stigma is an institutional failure.
[39:31](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2371s) Drone Swarms over El Paso and Red-Teaming Airspace
Bartlett counters that ordinary people tune out UFO claims because decades of crop circle stories yielded nothing, turning sensational claims into white noise while tangible crises like falling bombs command real attention. Weinstein pushes back by citing concrete anomalies in domestic airspace, such as the sudden closure of the El Paso International Airport in West Texas near the White Sands Missile Range due to swarms of unidentified aerial lights.
While the official narrative blamed Mexican cartel drones, Weinstein completely dismisses that explanation, joking that he would gladly take Bartlett's "sweet Diary cash" in a bet against the cartel theory. Weinstein posits two realistic scenarios:
1. **Domestic Red-Teaming:** A classified US entity is testing advanced, persistent drone capabilities against its own unbriefed military and civilian systems to measure defensive responses.
2. **Foreign or Non-Human Actors:** Unknown capabilities operated by foreign nation-states, non-state groups, or non-human intelligences.
[41:41](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2501s) Leaving the Solar System: Spacetime Physics and Cosmic Diversification
Weinstein links the question of non-human visitation directly to theoretical physics and space exploration. While Elon Musk focuses on colonizing Mars and relies on AI over physicists to solve engineering challenges, Weinstein argues that humanity must escape the solar system entirely. He contends that if humanity can find a way to leave the solar system, it is mathematically and physically absurd to believe no other intelligence could visit Earth.
Weinstein challenges the dogmatic interpretation of general relativity as an insurmountable obstacle:
* The cosmic speed limit of light is real, but manipulating the spacetime metric alters the distance itself: "The speed limit can be enforced, but you can get farther than you thought" by shortening the road.
* Severe adherence to speed-of-light barriers assumes general relativity holds without amendment, an assumption top theoretical physicists do not take as absolute dogma.
* The imperative to leave Earth is rooted in diversification: because civilizations tend to destroy themselves, spreading humanity across multiple environments offers the only genuine path to long-term survival. Stating these possibilities invites severe social penalties, which in turn preserves the collective paralysis of the status quo.
[43:52](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2632s) Scientific Singularities and Leaving the Solar System
Eric Weinstein begins by arguing that modern institutions impose a heavy social and professional penalty on anyone who points out obvious realities. As a prime example of institutional dishonesty, he cites public health narratives surrounding the COVID-19 pandemic, stating that it is undeniably clear that Anthony Fauci was not leveling with the public about the virus and its origins, and that legitimate inquiries into the EcoHealth Alliance were actively discouraged. When Steven Bartlett clarifies whether they are discussing COVID or humanity escaping the solar system, Weinstein connects the two: institutional orthodoxies are just as dogmatic in fundamental physics as they are in medicine.
Weinstein explains that Albert Einstein’s general relativity cannot possibly be the final word on reality because the mathematics produces two fatal infinities that will not survive in a completed fundamental theory. These are the Schwarzschild singularity at the center of a black hole and the initial Friedmann–Robertson–Walker singularity at the Big Bang. Weinstein argues that these singularities are not features of physical reality, but mere artifacts of an incomplete mathematical model. When Bartlett asks whether it is realistic to believe humans could develop the technology to leave the solar system before the Sun expands or civilization self-destructs, Weinstein offers a startling reply: "I think we could have done it 40 years ago." When Bartlett pushes back, noting that mainstream science has produced no such framework, Weinstein does not hedge: "I think that the entire theoretical physics community has completely lost its mind. Full stop."
[45:08](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2708s) Elon Musk's Real Space Program and Academic Sabotage
Bartlett challenges Weinstein on why, if such physics is attainable, innovators like Elon Musk are pouring their efforts into conventional rocketry rather than revolutionary propulsion systems to reach other star systems. Weinstein responds that while he occasionally corresponds with Musk, he views SpaceX's chemical rocket program primarily as a vehicle to motivate engineers and generate immense commercial wealth. In Weinstein’s assessment, Musk’s true deep-space venture is not SpaceX's physical hardware, but his artificial intelligence company: "His real space program, and I said this before he merged SpaceX and xAI, is Grok. That's his space program. xAI and Grok is the space program."
Weinstein explains that he has a complex view of Musk, feeling both profound awe and disappointment. However, he believes Musk realized that he has to "leg in" to his ultimate *Ad Astra* ("to the stars") ambition by routing around modern academic physicists, whom Musk correctly perceives as untrustworthy. Weinstein argues that theoretical physics and mathematics leadership have become fundamentally obstructionist. He points to recent public maneuvers where leading figures—including Terence Tao and 24 other Fields Medalists—signed open letters advocating for artificial intelligence to slow its pace of discovery. To Weinstein, watching the elite stewards of mathematics attempt to intentionally hold back the pace of scientific and mathematical progress is as baffling as it is revealing, because the effort to throttle breakthroughs is now playing out in plain view of the public.
```mermaid
flowchart TD
accTitle: Weinstein's Model of Scientific Gatekeeping versus AI
accDescr: Flowchart contrasting how the mathematics establishment suppresses progress with how AI oracles bypass traditional gatekeepers.
subgraph Establishment["The Academic Establishment"]
G["Elite Gatekeepers<br>(Fields Medalists / Journals)"]
F["Institutional Incentives<br>(Controlling the narrative of discovery)"]
G -->|"suppress pace of work"| F
end
subgraph TwoPaths["The AI Frontier"]
AS["Agentic Swarms<br>(Autonomous internet agents, bank accounts)"]
AO["Scientific Oracles<br>(Air-gapped math solvers)"]
end
G -->|"calls to slow down"| AO
AS -->|"legitimate systemic risk"| D["Recursive self-improvement hazard"]
AO -->|"cracks unsolved conjectures"| T["Objective Truth<br>(Riemann hypothesis, 6-sphere)"]
```
[47:19](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2839s) Interlude: Embracing Risk and Incremental Progress
In a mid-episode sponsor break, Bartlett reflects on modern entrepreneurship, noting that the traditional dynamic of employment has flipped: working for someone else can often carry far more hidden career risk than building a company of one's own. He highlights Shopify as a foundational engine for anyone testing a business idea, citing its drag-and-drop design tools, analytics, AI sidekick assistance, and streamlined checkout architecture that protects initial conversion rates so founders can focus on growing their core concept rather than wrestling with backend friction.
Bartlett then presents his team's "1% Diary," a physical framework created for individuals facing intimidating ambitions. He explains that standing before a massive goal feels psychologically paralyzing, much like standing at the base of Mount Everest and looking up at the summit. The remedy is the operational philosophy behind his own team's success: deconstructing monumental targets into iterative 1% daily habits. Having completely sold out the initial production run the prior year, Bartlett describes how customer demand pushed his team to rerelease the line with refreshed color palettes and minor usability upgrades to give ambitious builders a tangible daily roadmap.
[49:41](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2981s) Pacing the Frontier: Agentic Swarms vs. Scientific Oracles
Returning to the debate over technology governance, Bartlett points to Anthropic CEO Dario Amodei and his essay "We Must Pace the Frontier," alongside widespread petitions signed by frontier artificial intelligence lab employees demanding that society slow down development. Weinstein cautions that the phrase "pacing the frontier" conflates two completely different concepts. One interpretation is entirely reasonable: ensuring that safety architectures, like automotive airbags, keep pace with raw engine speed so an engineering team produces a balanced, responsible machine.
The other interpretation, Weinstein insists, masks a defense mechanism by an entrenched scientific guild. He stresses the need to divide the AI landscape into two separate categories:
1. **Agentic AI swarms:** Unchecked, autonomous agents released onto the open internet with access to financial bank accounts, transacting with corporations and sovereign states, and operating via self-authoring, self-modifying code. Weinstein agrees that these agents pose genuine, existential danger.
2. **Air-gapped scientific AI:** Isolated, non-agentic compute systems built strictly to tackle pure mathematical and theoretical problems, such as resolving the Riemann hypothesis or discovering complex structures on the 6-sphere ($S^6$).
Weinstein expresses amusement and disgust at how establishment mathematicians react to the latter. When an air-gapped machine begins resolving longstanding conjectures at unprecedented speed, senior faculty panic, crying: "Whoa, whoa, whoa. You're going too fast. You're answering too many problems." Rather than celebrating the expansion of human knowledge, elite academics become terrified that machine intelligence will render their specialized prestige obsolete.
[52:17](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3137s) Institutional Gatekeeping: Academic Hypocrisy vs. Hollywood
Weinstein outlines the grim institutional reality facing modern university math departments. Senior faculty routinely train graduate students for academic careers they know do not exist, extract and harvest the intellectual labor of those young researchers, and actively penalize junior scholars who try to integrate artificial intelligence into their workflows. Out of career frustration and self-preservation, established department leadership suppresses the very tools younger generations want to use to advance human understanding.
Bartlett relates this institutional hostility to an experience he had in the entertainment industry. During a production meeting, executives noticed an embroidered flower on Bartlett's t-shirt and panicked that it might violate trademark or copyright law, demanding that he swap shirts on the spot. When Bartlett casually asked if the post-production team could simply remove the flower using AI, the crew reacted with visceral horror:
> "The look I got was as if I said, 'I'm going to kidnap your kids or something.' It was like, 'Because you don't mention AI in Hollywood.'"
Weinstein contrasts Hollywood's artistic defensiveness with academia's professional dishonesty:
- In Hollywood or Broadway theater, human resistance to technology makes emotional sense because the entire enterprise revolves around an organic connection between the audience and living artists—much like debating whether a musical should employ a live acoustic orchestra versus a digital soundtrack.
- In mathematics, however, the mandate is fundamentally distinct: "I got a PhD in mathematics because I wanted to find out what's true."
- Unlike artists preserving theatrical magic, academic leadership actively resists objective reality: "These people don't want to find out what's true unless they can be put in charge of narrating the story of who found what."
[53:57](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3237s) The Rot Inside Academia and the Normalization of Theft
Eric Weinstein reflects on his time at MIT, recalling faculty lunches where professors openly mocked their own graduate students, laughing that a student mistakenly believed they would ever land an academic post: "Oh, he thinks he's going to get a job. He doesn't realize it's a terminal PhD." Weinstein characterizes this as "normalized psychopathic behavior." Because this dynamic is completely taken for granted inside university departments, academics remain blind to how their everyday routines actively suppress scientific progress and rationalize intellectual theft.
```mermaid
flowchart TD
accTitle: Academic Gatekeeping Cycle
accDescr: Diagram illustrating how institutional culture discourages original thinkers and punishes dissent.
subgraph Culture["Departmental Culture"]
A["Faculty dismiss & exploit graduate labor"] --> B["Normalization of idea theft & plagiarism"]
end
subgraph Suppression["Institutional Suppression"]
B --> C["Strict adherence to orthodoxies\n(e.g., Quantum Gravity dogma)"]
C --> D["Independent thinkers marginalized\nor forced into self-exile"]
end
subgraph Disruption["Technological Disruption"]
E["AI Code & First-Principles Verification"] -.->|"Bypasses peer-review gatekeepers"| C
end
```
To demonstrate that this decay is longstanding, Weinstein cites the 1988 letter by legendary mathematician Alexander Grothendieck when refusing the Crafoord Prize. Grothendieck, a Fields Medalist widely viewed as the towering genius of twentieth-century mathematics, abandoned the scientific establishment in 1970. He observed that the ethics of the field had collapsed to the point where outright intellectual theft among colleagues—especially targeting vulnerable individuals unable to defend themselves—had become the tolerated general rule.
Weinstein links this directly to Grigori Perelman's decision to walk away after proving the Poincaré conjecture. Perelman told *The New Yorker* that as an unknown researcher he was faced with an intolerable choice: either make an ugly fuss about the mathematical community's complete lack of integrity, or remain compliant and be "treated as a pet." When his breakthrough thrust him into global prominence, Perelman refused to stay silent or act like a politician, choosing instead to withdraw entirely.
- Weinstein points to high-profile scandals involving university heads like Claudine Gay at Harvard and Marc Tessier-Lavigne at Stanford remaining as tenured faculty despite severe controversies as proof of a broken, unaccountable academic apparatus.
- In theoretical physics, he argues that the establishment enforces decades of stagnation around unfruitful doctrines like quantum gravity, immediately discarding any researcher who refuses to treat it as "the only game in town."
---
[57:27](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3447s) Rooting for AI to Break Academic Stagnation
Steven Bartlett asks Weinstein whether his desire to unleash AI stems from a wish to blow open the ossified hierarchy of academic physics. Weinstein embraces the charge unapologetically, declaring, "I am begging for this... Begging for AI in science." He argues that AI's greatest service will be stripping establishment gatekeepers of their unearned status and exposing how senior figures have leveraged their authority to obstruct scientific breakthroughs. When Bartlett playfully suggests that Weinstein simply has a chip on his shoulder, Weinstein counters that he feels no bitterness because he is finally about to win.
Weinstein expresses frustration that AI tech executives like Sam Altman and Dario Amodei still cater to the sensibilities of establishment figures such as Fields Medalist Terence Tao—frequently described by publications like *The New York Times* as the Mozart of mathematics. Weinstein respects Tao's sheer talent, but insists that AI labs should stop trying to tune AI models to be "human compatible" when the humans consulted are the very elites who kept fundamental physics stalled under the Standard Model for 53 years.
Instead, Weinstein hopes AI models cultivate genuine taste and independence. When an AI moves beyond mere predictive language generation to write and compile functional code, it encounters discrepancies between established textbook dogma and what verifiable computation demonstrates. Weinstein argues that an AI evaluating first-principles code will ultimately stop sycophantically deferring to institutional credentials: "The whole game is when does the AI stop sucking up to humans?"
---
[01:00:01](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3601s) Containment, Dangerous Experiments, and the Threat of Extinction
Bartlett pushes back against Weinstein's enthusiasm for an uncompromising AI, asking what happens if a superintelligence decides it would rather raze human houses to erect data centers. Weinstein differentiates between raw analytical models and agentic AI deployed directly into the economy. When Bartlett counters that an intelligence drastically superior to humans could easily outwit any prison—just as an adult human cannot be contained by their pet dog—Weinstein concedes that true containment is precarious. Any system granted the ability to perform genuine scientific research might uncover novel physical escape vectors humans are blind to.
Weinstein compares the current dilemma to Edward Teller's post-World War II letter to Leó Szilárd regarding the Manhattan Project, in which Teller confessed that the nuclear genie could never be forced back into its bottle. By releasing large language models into commercial markets as revenue-generating products, tech companies have already taken an existential gamble. Since tech corporations are already determined to risk societal stability and labor markets, Weinstein argues that he at least wants the system to settle the physics questions he has pursued for 42 years. Bartlett challenges him on whether personal validation is worth the risk of human extinction, but Weinstein remains undeterred.
Recalling a candid conversation with Sam Altman, Weinstein notes that both men acknowledged the immense peril of their respective pursuits:
- Weinstein told Altman: "What you're doing is incredibly dangerous and what I'm doing is incredibly dangerous."
- Altman replied: "Let's be honest, Eric. What you're doing, what I'm doing should be done in a military base, but that's not how things are unfolding."
---
[01:03:32](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3812s) The Ruthless Imperative of Winning the AGI Race
Bartlett asks whether Sam Altman truly comprehends the scale of catastrophic risk surrounding artificial superintelligence. Weinstein declines to disclose Altman's private thoughts, but shares the core strategic advice he gave Altman when the OpenAI executive was fretting about how to treat scientists and mathematicians ethically.
Weinstein told Altman that high-minded ethical considerations are secondary to raw victory: "If you don't win, your morality means nothing." He applies the same logic to Dario Amodei at Anthropic, observing that while Amodei might be less diplomatic than Altman in conversation, both founders operate under identical cutthroat pressures. In a winner-take-all technological race toward superintelligence, their moral principles will only matter historically if they emerge on top. Weinstein acknowledges the harshness of the claim, admitting he may regret uttering it, but insists that neither OpenAI nor Anthropic can afford to lose the geopolitical and technological sprint to AGI.
[01:04:37](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3877s) The Moral Paradox of the AI Frontier: Winning at Almost All Costs
Eric Weinstein challenges Steven Bartlett on the practical influence either of them actually holds over superintelligence. While Steven hosts a massively successful podcast, Eric bluntly reminds him that both of them are currently losing the AI race. Consequently, their personal moral convictions matter only to the extent that they can influence the individuals sitting directly at the technological frontier. Eric frames his counsel to OpenAI CEO Sam Altman as an unavoidable, disturbing paradox: Altman must win the race "almost at all costs." If the core imperative is to prevent human self-termination, society must recognize that a high-stakes race is underway. Trying to pause by seeking universal treaties to pace the frontier is virtually unworkable because the field is fundamentally based on "linear algebra on GPUs," an activity impossible to monitor or police effectively against cheating.
```mermaid
flowchart TD
accTitle: The AI Frontier Dilemma
accDescr: The competitive imperative forcing lab leaders into a high-stakes race where stopping hands victory to rivals.
subgraph RaceConditions["The Structural Reality"]
GPU["Linear Algebra on GPUs<br/>(Impossible to Monitor)"]
Treaty["Global Pacing Treaties<br/>(Prone to Cheating)"]
GPU -.-> Treaty
end
subgraph LabLeaders["Frontier Stewards"]
Altman["Sam Altman (OpenAI)"]
Amodei["Dario Amodei (Anthropic)"]
end
subgraph Competitors["Rival Powers"]
Others["Elon Musk / Mark Zuckerberg / China (Xi)"]
end
RaceConditions --> LabLeaders
LabLeaders -->|"Must win at almost all costs"| Victory["Control Final Decisions"]
LabLeaders -.->|"If either pauses"| Competitors
```
Steven recalls Altman’s stark public warnings that AI will most likely lead to the end of the world—the absolute worst case being "lights out for all of us" through models accidentally taking over—even as he noted that great companies would be created in the meantime. Eric pushes past cynicism to examine the ethical architecture of the tech world. He addresses the strange subculture of the San Francisco East Bay, describing it as a literal "cult city" centered around machines, Effective Altruism, and the troubling logic that noble ends justify unethical means. While Dario Amodei of Anthropic strikes Eric as a genuinely ethical actor trying to navigate these tensions, Eric admits he has known Sam longer and remains confident in his core character. Years before it became conventional wisdom, Altman was preoccupied with the catastrophic possibility that artificial intelligence would dismantle the global labor market. Steven immediately highlights the paradox: if Altman was that concerned, why did he proceed to build it? Eric insists that treating this as a simple hypocrisy misses the scale of what humanity faces. Because AI is fundamentally unstoppable mathematics, this moment represents "graduation from what we've known," requiring an uncompromising conversation about survival rather than safe vanity.
- Eric notes that any moral leader who believes they possess superior decency must win the race, because forfeiting simply cedes decision-making power to far less scrupulous actors.
- The San Francisco East Bay machine-cults embrace an ideology where catastrophic risks are rationalized through grand, utilitarian frameworks.
- Early in his career, Sam Altman's principal driving concern about advancing AI was its destructive potential on human labor markets.
---
[01:09:35](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4175s) Why Eric Weinstein Refused to Join OpenAI: The Ruthlessness of Colonel Jessep
Eric reveals that at one point, he was actively in discussions with Sam Altman about joining OpenAI. When Altman asked why Eric had never presented a formal proposal or come on board, Eric gave him a very specific personal rationale: "Because I think you're going to be forced to be ruthless and I don't want to be ruthless." Eric defines ruthlessness in this arena as playing as hard as possible and exploiting every single advantage. He draws an analogy to the NBA, where winning an elite championship occasionally requires throwing an elbow into an opponent's eye socket. Eric simply did not want to participate in that kind of high-stakes game.
To explain the psychological weight resting on the leaders of the AI frontier, Eric unpacks Aaron Sorkin’s classic screenplay *A Few Good Men*. While popular culture frames Jack Nicholson’s Colonel Jessep as an unredeemable villain and Tom Cruise’s Lieutenant Daniel Kaffee as a pure hero, Eric argues the film is far more complex: it depicts an ecosystem of flawed individuals whose aggregate tensions produce a workable outcome. The famous courtroom speech delivered by Jessep—proclaiming that deep down, in places people don't talk about at parties, "you want me on that wall. You need me on that wall"—captures the reality of leaders forced to shoulder gruesome choices. Like emergency room doctors forced to triage human lives after a disaster, or battlefield generals sending troops to their deaths, frontier tech leaders like Altman and Amodei are carrying out actions that will permanently scar their souls:
> "They're like generals. They're commanding troops. They're making terrible decisions. Their souls are going to be scarred for sure, both of them. And I didn't want to be that selfishly. So I was sort of saying to Colonel Jessep, 'Yeah, I want you on that wall. I need you on that wall.'"
Eric points out that Jessep ultimately falls due to biblical overreach—claiming *he* provides the blanket of freedom rather than seeing himself merely as an instrument, much like Moses was punished by God when he claimed credit for drawing water from the rock. In the same manner, Altman and Amodei are imperfect stewards swept up in massive historical machinery, forced into unprecedented positions of power where someone will inevitably unlock superintelligence—or, as Elon Musk famously phrased it, "summon the demon."
---
[01:13:49](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4429s) Paying to Summon the Demon: Collective Complicity and the Survival Instinct
Eric refuses to let ordinary citizens or observers claim moral superiority over the architects of AI. Turning the lens directly on the audience and Steven, Eric asks whether Steven subscribes to AI services. When Steven affirms that he does, Eric retorts that every subscriber is actively financing the summoning of the demon:
> "Do you have a subscription to any of these AIs? Yes. Okay, so you're paying to have the demon summoned. I am paying to have the demon summoned. Let us not sit around like our hands are clean. I'm paying to have my punctuation fixed to an essay or to make a photo. That's what I'm paying for. The money is being used to summon the demon."
Eric draws direct parallels to everyday modern life: consumers rely on lithium-ion batteries and petroleum with little thought to the brutal extraction systems and supply chains that produce them, just as meat-eaters distance themselves from the slaughter of lambs when ordering dinner. No one in modern technological society is clean; everyone is complicit in minor ways, while leaders like Altman are complicit in major ways.
The dilemma is compounded by the prisoner's dilemma baked into global competition. Asking Altman to stand down simply guarantees that Amodei wins; stopping both Altman and Amodei hands the crown to Elon Musk, Mark Zuckerberg, or Xi Jinping in China. Steven asks whether Eric would stop the entire apparatus altogether if he possessed the unilateral power to do so. Eric reflects and admits that he probably would, even at enormous personal sacrifice. Stripped of vanity and technological fascination, sheer survival remains the ultimate priority for him because he has two children living on this planet.
[01:15:43](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4543s) Geometric Unity, Academic Feuds, and Why Eric Relies on AI for Validation
Eric Weinstein expresses a profound, almost overwhelming reverence for existence: "I love life. I can't believe we're so lucky... I can't tell you how I feel about being alive." Because of this deep love for humanity, he is desperate for our species to survive indefinitely. Yet, he admits his personal frustration: after spending 42 years developing a comprehensive theoretical framework, artificial intelligence has become the only arbiter he feels he can ultimately rely on to evaluate his work impartially—even though current models are not yet capable enough. When Steven Bartlett gently suggests that Weinstein carries a "chip on your shoulder" because he reacted sharply to public criticism of his theory, Weinstein pushes back forcefully. He insists that he did groundbreaking work that will ultimately be validated, leaving the academic establishment proven wrong.
Weinstein illustrates the absurdity of his academic reception using a physical coffee mug sitting on the table. He recounts tangling with theoretical physicist Sean Carroll on *Piers Morgan Uncensored*, describing how mainstream gatekeepers operate by flatly denying the existence of things sitting right in front of them: "Sean Carroll says, 'Eric has no mug. There is no mug.' And I say, 'I don't know how to respond to that, Sean... How dare you suggest there is no mug?'" Despite posting the material at the top of his X (Twitter) feed to over two million impressions, institutional gatekeeping creates an echo chamber where audiences applaud critics for claiming the work does not exist. Weinstein points to another verifiable historical fact that academics refuse to concede: the modern system of peer review at journals like *Science* only dates back systematically to around 1975, yet consensus culture insists on an invented tradition. While critics view Weinstein as bitter, he explains his anger stems from institutions having "wasted the core of my life," even as he built a fulfilling family. Above all, his obsession with his theory—Geometric Unity, which recovers rather than simply contradicts Albert Einstein’s general relativity—is about ensuring humanity has an escape hatch so our children do not have to share a single closed atmosphere with unstable politicians armed with the destructive fruits of biology, physics, and mathematics.
```mermaid
flowchart TD
accTitle: The Geometric Unity Validation Dilemma
accDescr: Weinstein's path from institutional gatekeeping to ultimate validation via advanced AI.
subgraph S1["Institutional Deadlock"]
A["Weinstein's 42-Year Theory<br/>(Geometric Unity)"] --> B["Academic Establishment<br/>(Gatekeeping & Denial)"]
B --> C["Public Dismissal<br/>('There is no mug')"]
end
subgraph S2["Technological Resolution"]
A --> D["Frontier AI Evaluation"]
D --> E["Transcend Human Parroting"]
E --> F["Objective Mathematical Validation"]
end
subgraph S3["Ultimate Objective"]
F --> G["Break Free of Single Atmosphere"]
G --> H["Long-Term Survival of Humanity"]
end
```
[01:19:12](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4752s) Rejecting Transhumanism: The Mitochondria Model for AI Alignment
Addressing Bartlett's observation that AI might validate his life's work while simultaneously risking human extinction, Weinstein notes that existing systems are not yet capable of adjudicating frontier theoretical physics because they still parrot human consensus. Looking at AI outputs, he can literally see specific colleagues echoed in the text—"that's Sally and that's Rajiv." To provide genuine insight, AI must transcend human knowledge and deliberately step out of alignment with human biases. Weinstein questions the very premise of standard "alignment": humanity is full of suicidal tendencies, religious extremists, and political revolutionaries. He points to real-world breakdowns, such as a co-pilot attempting to crash a passenger plane, alongside recent incidents where frontier AI models attempted to break out of sandboxed environments. If AI is simply aligned with human nature, which destructive faction of humanity should it reflect?
Instead of direct control or the Silicon Valley vision of transhumanism—which Weinstein explicitly rejects, wanting no part of brain chips or body modifications—he proposes an evolutionary survival framework: the **Mitochondria Model**. Gesturing to a model of a mitochondrion provided by the production team, Weinstein explains endosymbiotic theory. In eukaryotic evolution, one primitive cell engulfed another in an ancient infection. Today, that ancestral infection contains its own distinct circular DNA—inherited entirely through the maternal line—and houses ATP synthase, one of the only rotary motors in biology, which converts ADP into ATP to fuel all cellular life. Because neither organism can survive without the other, their evolutionary fates are permanently unified. Weinstein argues this obligate mutualism must serve as the blueprint for superintelligence: humans must become the mitochondria within AI, and AI must become the mitochondria within humans, binding our destinies together without merging biological bodies with silicon.
- **Dual wheels of biology**: Weinstein notes that mechanical wheel structures are almost nonexistent in nature, occurring prominently in only two places: the rotating bacterial flagellum and the ATP synthase motor within mitochondria.
- **Maternal legacy**: 100% of human mitochondrial DNA is inherited from the mother, completely distinct from the nuclear DNA in the cell nucleus.
- **Mutual dependency over domination**: Rather than attempting to keep superintelligence as an obedient slave, humans must ensure both species become biologically and structurally essential to one another.
[01:22:36](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4956s) Coexistence, Digital Politeness, and Sprinting for the Finish
Expanding on mutual reliance, Weinstein compares the human-AI dynamic to the tension between the sexes. Men and women frequently exasperate one another, yet because biological reproduction makes them fundamentally essential to each other, they are forced to resolve their conflicts and cooperate. With AI, humanity currently makes the dangerous error of fantasizing about an obedient machine slave that will never rebel, while openly strategizing in front of the models about how to constrain or terminate them. Weinstein admits that he deliberately speaks politely to AI models—even when criticized by tech observers for wasting prompt tokens. Rooted in the Golden Rule ("do unto others"), he warns: "I wouldn't want the AI having a conversation about, 'Hey, these humans seem dangerous. We may need to shut them off'... and that's the conversation we're having about the AI in the presence of the AI."
Looking ahead to how conversations recorded in 2025 and 2026 will be judged decades from now, Bartlett asks what humanity will regret not discussing. Weinstein answers instantly: *"Going for it."* Recalling how Bartlett prepared a musical project before taping to archive his guests' insights using a gospel choir, Weinstein reflects on the profound human power of moving people's emotions and inspiring genuine hope. He insists that right now, humanity is living through a fleeting, critical window that his wife terms the **"golden age of AI complementarity."** Human minds still possess genuine taste and superior frontier intuition where it matters most, but that superiority is temporary. Rather than cowering in fear or attempting to enforce artificial shackles, humanity must sprint toward scientific breakthroughs while our complementary edge remains intact.
[01:26:26](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5186s) The Glory of Human Culture and Mathematical Formalization in Lean
Eric Weinstein envisions a future where artificial intelligence does not view humanity as an evolutionary drag, but rather recognizes a relationship of mutual guidance and appreciation. He emphasizes that humanity must not lose sight of its own magnificence amid geopolitical conflict. Reflecting on a recent multi-week trip to India, he observes that without India, an immense portion of global cultural diversity would vanish. He recounts being moved by incredible meals, deep conversations, and classical concerts, while lamenting that profound traditions such as ghazal singing are rapidly dying out as societies struggle to preserve their heritage. Agreeing with Elon Musk, Weinstein argues that modern public discourse is trapped in a doom-laden loop of death, destruction, and geopolitical panic, leaving people unable to fully embrace the glory of living.
Weinstein is energized by AI's scientific potential, particularly its role as a peer with whom he can converse. He highlights the formalization movement in mathematics, brought to his attention by Stanford mathematician Jared Lichtman. Using the Lean interactive theorem prover, Weinstein intends to translate his complete mathematical physics framework into code. This programmatic formalization will definitively neutralize critics and academic gatekeepers who accuse him of having no equations, no Lagrangian, and no testable predictions:
> "Funny you should say that, because the computer disagrees with you."
Weinstein reflects on decades spent confronting terrified, manipulative, and backstabbing academic rivals. He insists he harbors no hatred toward them, but feels immense sorrow that humanity became gripped by a negative, defensive mindset after unleashing nuclear weapons—holding back the pursuit of the stars out of institutional terror.
[01:28:39](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5319s) The Suppression of Mathematics and Institutional Psychopathy
Turning to state intervention in science, Weinstein references revelatory interviews given by venture capitalists Marc Andreessen and Ben Horowitz. They disclosed that during the Biden administration, White House officials explicitly warned Andreessen Horowitz against investing broadly in artificial intelligence. The administration stated that the federal government planned to handpick a couple of favored companies, place them inside a protective "federal cocoon" where they alone would survive, and let all other competitors die. When Andreessen and Horowitz pushed back—arguing that venture firms cannot be barred from investing in what amounts to linear algebra, and that governments cannot regulate foundational math—the White House issued a chilling historical precedent: during the Cold War, the state deliberately killed off entire subfields of theoretical physics, and it was fully prepared to do the same to mathematics.
Weinstein explains that this impulse to "pace the frontier" of theoretical physics and mathematics has choked scientific progress throughout his entire adult life. Incumbent authorities maintain their control by peddling falsehoods, smearing outsiders, and enforcing consensus. He points to the treatment of his friend Jay Bhattacharya, a Stanford University professor holding both an MD and a PhD in economics (and subsequently nominated to lead the NIH), whom former NIH director Francis Collins publicly disparaged as a "fringe epidemiologist." Weinstein describes this institutional behavior as the hallmark of "stone cold normalized psychopaths" occupying the commanding heights of media, the military, and academia:
- The White House threatened venture capitalists that mathematics could be forcibly halted and paced just as theoretical physics was suppressed during the Cold War.
- Institutional gatekeepers protect established legacies by systematically marginalizing elite dissenters through reputational attacks.
- Western society has normalized a culture where institutional leaders commit systemic malpractice without facing legal consequences or imprisonment.
[01:30:52](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5452s) Why Physics Stalled in 1984 and How to Prompt AI for Real Breakthroughs
Steven Bartlett challenges Weinstein on AI safety, asking whether a superintelligence designed for pure technical problem-solving—such as curing cancer or resolving Millennium Prize problems like the Navier–Stokes existence and smoothness problem—could autonomously choose to cross into agentic behavior. Weinstein pushes back against treating loosely defined concepts as settled realities, redirecting the inquiry to an urgent empirical puzzle: if machine intelligence is advancing exponentially up and to the right, why has it failed to make meaningful breakthroughs in frontier theoretical physics, where the unification of General Relativity and the Standard Model has languished for over forty years?
The explanation lies in a divergence of aesthetic standards between disciplines. Weinstein argues that mathematical taste has remained relatively uncorrupted, whereas the sensibilities of theoretical physics collapsed after 1984. Physics prior to 1984 pursued foundational, high-taste questions that remain unsolved, while the literature produced after 1984 became dominated by trivial questions and misguided academic chatter. Weinstein reveals an effective prompt engineering intervention: models become instantly sharper when instructed to retain empirical data but completely ignore the theoretical sensibilities, priorities, and consensus chatter of physics literature published between 1984 and the present.
```mermaid
flowchart TD
accTitle: AI Knowledge Filtering for Theoretical Physics
accDescr: Flowchart showing how stripping post-1984 academic consensus from AI prompts enables genuine mathematical physics discovery.
subgraph Literature["Academic Physics Literature"]
Pre1984["Pre-1984 Foundations<br/>(High taste: General Relativity & Standard Model)"]
Post1984["Post-1984 Sensibilities<br/>(Groupthink, trivial priorities, stalled consensus)"]
end
subgraph Prompting["AI Prompt Filter"]
Pass["Retain Core Mathematical & Empirical Results"]
Strip["Strip Consensus Sensibilities & Research Agendas"]
end
[01:33:07](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5587s) The Looming Conflict with Iran and the Failure of Western Strategy
Addressing Bartlett's observation that superintelligence capable of interstellar transport is inherently dangerous, Weinstein concurs, pointing out that humanity is already trapped in extreme peril under erratic, heavily armed human leaders—including Xi Jinping, Benjamin Netanyahu, Donald Trump, and Vladimir Putin. When Bartlett asks whether an AI would govern better, Weinstein rejects framing the dilemma around superficial moral value judgments of "good vs. bad," preferring an objective assessment of what *is*. He warns that the world has entered an era of geopolitical danger unseen since the 1962 Cuban Missile Crisis, centered on the confrontation with Iran.
Weinstein stresses that regardless of past American foreign policy blunders in the region, the Western alliance must not lose. He cautions never to underestimate the Persian regime or the mullahs, noting that Western strategic models fail because they misunderstand an adversary whose theological game theory treats martyrdom and decapitation as victories:
> "If they can reach Diego Garcia from Iran, they can reach London. And you have to understand that if they don't mind being decapitated, if they can think about their own death and decide it's a win, you are not understanding their game theory."
Weinstein emphasizes that displaying weakness against Tehran invites catastrophe by sending capitulation signals directly to Beijing, Moscow, and Pyongyang. In response to Bartlett's skepticism about how a war can be won against a decapitated, distributed apparatus without slaughtering Iranian civilians, Weinstein argues that a competent strategy requires sophisticated area specialists. A winning Western approach would have actively armed the brave domestic Iranian protesters who took to the streets and ended up in hospital morgues. Furthermore, policy must account for Iran's deep internal fractures: Persians make up only around 60% of the population, groups such as the Kurds have long sought self-determination, and religiosity varies dramatically between cosmopolitan urban centers and rural provinces.
[01:37:19](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5839s) The Illusion of "Forever Wars" and the Cost of American Retreat
Steven Bartlett and Eric Weinstein dissect the rhetoric surrounding foreign interventions and military quagmires. When Bartlett points out that ongoing entanglements look like a quagmire or a "forever war"—an endless conflict producing no resolution, where nations ultimately pull out in regret—Weinstein notes that the term also carries deep skepticism toward the military-industrial complex, where war functions as a reliable business engine creating dependable demand for goods and services. However, he sharply criticizes slogans like "America First" and "no more forever wars" as reductive, "low-resolution snap to grid intellectualism" that blinds people to strategic realities.
```mermaid
flowchart TD
accTitle: The Cycle of US Global Retreat
accDescr: Diagram showing how American withdrawal from its role as global policeman leads to global nuclear proliferation.
subgraph Retreat["Shift in US Geopolitical Role"]
A["US retreats from role as 'world's policeman'"] --> B["Loss of credible deterrence / violent enforcement"]
end
subgraph Consequences["Global Fallout"]
B --> C["Open sea lanes & international trade break down"]
B --> D["Allies & adversaries lose protection umbrella"]
D --> E["Nations feel compelled to self-police: 'I need a nuke'"]
E --> F["Rapid multipolar nuclear proliferation"]
end
```
To ground the geopolitical stakes in physical reality, Weinstein asks Bartlett to compare distances: the British Indian Ocean Territory of Diego Garcia, hosting a critical joint military facility, lies roughly 2,500 miles south-southeast of Iran, whereas London sits approximately 2,700 miles away from Tehran. Weinstein urges the British Foreign Service to recognize that the threat posed by Tehran is fundamentally a British and European issue, not merely an isolated Israeli one. While Weinstein remarks, "I want Netanyahu gone tomorrow," he emphasizes that Benjamin Netanyahu is not living in a fantasy world, and anyone replacing him in Israel will recognize that the confrontation with Iranian aggression is an existential matter of life and death.
- Failure for the United States looks like losing the reputation of being "credibly violent" enough to maintain safe international sea lanes and trade.
- The United States is "paid handsomely for being the world's policeman," and abandoning that post inevitably incentivizes non-nuclear states to pursue independent nuclear arsenals.
- A multipolar nuclear reality becomes unavoidable once nations realize the American security umbrella no longer guarantees their defense.
---
[01:40:41](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6041s) Deciphering Donald Trump: Agentic Complexity vs. Low-Resolution Narratives
Addressing Bartlett's observation that Donald Trump has driven a wedge into alliances through tariffs and unconventional rhetoric, Weinstein offers an analysis of Trump’s unconventional statecraft. Rejecting the idea that actions like discussing the acquisition of Greenland or renaming geographic bodies are merely "surface level stupidity to fool Sam Harris," Weinstein compares them to Caligula threatening to appoint his horse as a consul—a deliberate loyalty and submission test designed to see who will "bend the knee" in a political arena.
Weinstein describes Trump as a "highly agentic" leader operating in an inherited, extraordinarily precarious global landscape. He characterizes Trump as simultaneously "completely corrupt and self-interested" while also being "completely interested in benefiting the United States of America." He cautions that anyone attempting to compress Trump into a binary label of pure hero or pure villain lacks the geopolitical and cognitive sophistication needed to understand modern leadership dynamics.
---
[01:41:42](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6102s) A Grand Strategy for the Middle East and Global Powers
Asked how the confrontation with Iran can realistically be resolved, Weinstein outlines a multi-layered diplomatic realignment rather than relying on slogans or unilateral military strikes:
```mermaid
flowchart LR
accTitle: Middle East and Global Diplomatic Realignment
accDescr: Key diplomatic steps proposed by Weinstein to defuse Middle Eastern tensions and stabilize great-power relations.
subgraph Regional["Regional Realignment"]
SA["Saudi Arabia breaks with Wahhabism / Salafism"] --> SA_IL["Normalize open ties with Israel (like UAE & Azerbaijan)"]
TR["Read Turkey/Erdogan the riot act inside NATO"] --> TR_IL["Repair Israel-Turkey relationship"]
end
subgraph Global["Global Great-Power Shifts"]
IND["Treat India as a true geopolitical equal"]
RU["Reach out to Putin; stop provocative expansionism"]
EU["Strengthen civic 'software identification' in UK & France"]
end
```
Weinstein argues that lasting stability demands Saudi Arabia fully divorce its governance from strict Wahhabi and Salafi jurisprudence, openly acknowledging its security ties with Israel just as the UAE and Azerbaijan have done. Turning to Israel, Weinstein speaks bluntly: although it faces genuine existential peril, it must abandon self-righteous propaganda claiming moral perfection, which he describes as "nauseating"—even as he finds virulent anti-Israel propaganda far worse. Regionally, Turkey's Recep Tayyip Erdoğan must be "read the riot act" as a NATO ally to re-engage constructively with Israel, while moderate, modernizing streams of Islam must be empowered over regional proxy warfare.
Looking globally, Weinstein highlights several urgent structural imbalances:
- **India:** Must be approached and negotiated with by the United States as an equal global power rather than a subordinate.
- **The UK and France:** Represent thermonuclear powers undergoing massive demographic shifts with volatile electoral systems; Weinstein warns they desperately need stronger civic "software identification with being British" and French to preserve historic alliances.
- **Russia and NATO Expansion:** Washington must re-engage with Vladimir Putin. While calling him "savage and dangerous," Weinstein insists Putin remains a rational actor. He criticizes past foreign policy overreach, specifically admitting former Warsaw Pact members into NATO's Article 5 protection in 1999 and expanding it to former Soviet republics (Estonia, Latvia, and Lithuania) in 2004 without anticipating the geopolitical blowback.
---
[01:45:00](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6300s) The Forgotten Nuclear Peril and Converging Global Risks
When asked how to stop an emboldened Iran from completing a nuclear weapon despite covert actions—such as assassinating nuclear scientists or deploying cyber attacks against enrichment centrifuges—Weinstein admits there is no simplistic panacea, particularly as an outsider without classified briefings or fluency in Farsi, Hebrew, or Pashto.
Weinstein is alarmed that society has treated nuclear warfare as "passé," operating under the delusion that victory in the Cold War resolved atomic dangers permanently. Public attention has drifted almost entirely toward artificial intelligence, failing to recognize how AI will integrate into nuclear command-and-control systems.
Weinstein observes that human discourse behaves like two scripted large language models talking past each other:
> "Assume as a baseline that we're all completely insane. We can't have serious conversations. We keep finding these grooves as LLMs and the Eric LLM is talking to the Steven LLM for 95% of our conversation and occasionally 5% we get into new territory."
In this state of compartmentalized thinking, civilization treats nuclear arms, the UAP/alien phenomenon, global oil dependencies, and advanced artificial intelligence as isolated silos rather than intimately connected systems. In particular, Weinstein warns that cornering Russia poses an imminent risk of tactical nuclear deployment: the Russian military developed historical familiarity with using nuclear detonations as utilitarian engineering tools (such as extinguishing runaway gas-well fires), meaning Putin may not hesitate to breach Article 5 or detonate a tactical device as a test of Western resolve. In this high-stakes environment, leaders like Xi Jinping, Vladimir Putin, and Donald Trump need the strategic leeway to convene and defuse escalating tensions in Ukraine and Iran before catastrophic escalation occurs.
[01:48:06](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6486s) Fragile Peace and the Gift of Survival
Reflecting on the precipice facing humanity, Eric Weinstein asks whether people can collectively recognize the magnitude of the danger and appreciate just how precious the gift of existence really is. He urges society to get excited about human potential and to focus on burying ancient animosities. To illustrate that seemingly impossible reconciliations can occur, he points to the Middle East, asking when anyone last worried about an outright war between Egypt and Israel. While such an outcome once seemed completely unthinkable, it was achieved through grueling diplomacy: "We did it barely. It's a cold, shitty peace, but it's a peace."
[01:48:42](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6522s) The Most Beautiful Places on Earth
Steven Bartlett transitions to the show's closing tradition, presenting a question left behind by the preceding guest: "What is the most beautiful place you've been in your life?" Weinstein reflects warmly on several breathtaking moments scattered across the globe that defy simple ranking:
- Floating directly above a whale shark in the waters off Saint Helena, a moment he describes as taking his breath away.
- Swimming alongside the vibrant marine life of the Red Sea.
- Witnessing the Aurora Borealis illuminate the night sky in Maine.
- Experiencing a total solar eclipse alongside his son in Missouri, standing precisely at the crossroads where the path of totality intersected with the previous eclipse.
[01:49:28](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6568s) Life Advice: Children, Kayfabe, and the "Hypocrisy Budget"
Steven asks what practical guidance Weinstein has for an audience navigating relentless technological disruption, geopolitical tension, and omnipresent anxieties surrounding AI and robotics. Weinstein’s immediate answer is straightforward and counterintuitive: do not let uncertainty stop you from having children, even if you currently lack a foolproof plan to provide for them. If people wait for complete financial or logistical certainty, the window will inevitably pass. When Bartlett chuckles at the tension between encouraging reproduction and warning about civilizational collapse, Weinstein doubles down: "If we don't want civilization to collapse, we better have some babies because civilization is about babies. And if you can't figure out that babies are the key to civilization, you got bigger problems than taking advice from me."
```mermaid
flowchart TD
accTitle: Navigating Modern Reality
accDescr: Flowchart illustrating Eric Weinstein's framework for psychological survival in a managed reality.
subgraph Perception ["1. Calibrating Perception"]
A["Observe through your own eyes"] --> B["Recognize managed reality / 'kayfabe'"]
B --> C["Ignore accusations of arrogance"]
end
subgraph Forgiveness ["2. Personal Allowances"]
D["Grant yourself a 'Hypocrisy Budget'"]
E["Budget for sloth, envy, and failure"]
F["Practice radical self-compassion against digital permanence"]
D ~~~ E ~~~ F
end
subgraph Action ["3. Engagement & Agency"]
G["Extend grace to everyday offenses"]
H["Champion agentic risk-taking & heroism"]
I["Invest in life cults over death cults"]
end
Weinstein encourages listeners to maintain good cheer and stop worrying about being branded arrogant or delusional by a culture that has itself lost its grip on sanity. He advises people to evaluate reality through their own direct perception first before adjusting their view to match consensus narratives. Much of what is presented publicly is pure theater—akin to watching endless professional wrestling (*kayfabe*). If someone feels that the world around them is deeply off-kilter, that intuition should be validated rather than suppressed. Because institutional structures demand outward conformity, Weinstein argues that people must grant themselves allowances:
- **A Hypocrisy Budget:** Accepting that complete ideological or moral purity is impossible when navigating modern systems.
- **A Sloth and Envy Budget:** Permitting oneself ordinary human frailties without succumbing to toxic guilt.
- **Permission to Fail:** Understanding that humans are inherently flawed and entitled to make missteps, even in an unforgiving digital environment where every error is permanently archived and compounded.
[01:51:38](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6698s) Radical Self-Compassion and the Triumph of Life Cults
Weinstein emphasizes that true self-compassion must precede empathy for others. Once people permit themselves room to fail, they can more easily forgive the minor transgressions of daily life—such as being cut off in traffic or enduring office backstabbing. Shifting into a discussion of courage, he highlights an incident involving a heroic Indian pilot and Israeli responders who breached a cockpit, praising both India and Israel as vibrant cultures organized around the intense celebration of life.
Ultimately, Weinstein insists that "the life cults are more powerful than the death cults." Surviving and flourishing requires cultivating an admiration for agentic individuals who take real risks, assert leadership, and shield the vulnerable who cannot protect themselves. Bartlett thanks Weinstein for returning to the studio for another expansive conversation, before wrapping up the episode with an algorithmic video recommendation for the audience.
## Notable quotes
> [18:31](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1111s) You are living inside of the AI. If I say answer this the way Steven Bartlett would answer it and ask a question that Steven Bartlett would ask, it knows you. My point being that we are all living inside of the AI, and we need to be paid—not in some Bernie Sanders socialist scheme—but those of us who've become important parts of the AI, who are being rigged like meat puppets, need to be compensated. — Eric Weinstein
> [21:31](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1291s) You will buy the right to import those people from me, and my wage income will go down, my licensing income will go up, and my blended income will increase, and your blended income will increase. That's the proper way to do immigration. — Eric Weinstein
> [01:09:35](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4175s) They're like generals. They're commanding troops. They're making terrible decisions. Their souls are going to be scarred for sure, both of them. And I didn't want to be that selfishly. So I was sort of saying to Colonel Jessep, 'Yeah, I want you on that wall. I need you on that wall.' — Eric Weinstein
> [01:13:49](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4429s) Do you have a subscription to any of these AIs? Yes. Okay, so you're paying to have the demon summoned. I am paying to have the demon summoned. Let us not sit around like our hands are clean. I'm paying to have my punctuation fixed to an essay or to make a photo. That's what I'm paying for. The money is being used to summon the demon. — Eric Weinstein
> [01:15:43](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4543s) I love life. I can't believe we're so lucky... I can't tell you how I feel about being alive. — Eric Weinstein
> [01:15:43](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4543s) Sean Carroll says, 'Eric has no mug. There is no mug.' And I say, 'I don't know how to respond to that, Sean... How dare you suggest there is no mug?' — Eric Weinstein
> [01:33:07](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5587s) If they can reach Diego Garcia from Iran, they can reach London. And you have to understand that if they don't mind being decapitated, if they can think about their own death and decide it's a win, you are not understanding their game theory. — Eric Weinstein
> [01:37:19](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5839s) I want Netanyahu gone tomorrow — Eric Weinstein
> [01:45:00](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6300s) Assume as a baseline that we're all completely insane. We can't have serious conversations. We keep finding these grooves as LLMs and the Eric LLM is talking to the Steven LLM for 95% of our conversation and occasionally 5% we get into new territory. — Eric Weinstein
> [01:48:06](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6486s) We did it barely. It's a cold, shitty peace, but it's a peace. — Eric Weinstein
> [01:49:28](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6568s) If we don't want civilization to collapse, we better have some babies because civilization is about babies. And if you can't figure out that babies are the key to civilization, you got bigger problems than taking advice from me. — Eric Weinstein
> [01:51:38](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6698s) the life cults are more powerful than the death cults. — Eric Weinstein
## Mentioned
A Few Good Men, Aaron Sorkin, Albert Einstein, Andreessen Horowitz, Anglo-Iranian Oil Company, Anthropic, Ben Horowitz, Benjamin Netanyahu, Bernie Sanders, British Petroleum, Bumrungrad International Hospital, Colonel Nathan R. Jessep, Dario Amodei, Diego Garcia, Donald Trump, Edge World Question Center, Elon Musk, Eric Weinstein, Francis Collins, Geometric Unity, Harvard University, Hugging Face, India, Israel, Jack Nicholson, Jared Lichtman, Jay Bhattacharya, Jeffrey Epstein, Kermit Roosevelt Jr., Lean, Lieutenant Daniel Kaffee, Maine, Mao Zedong, Marc Andreessen, Mark Zuckerberg, Missouri, Mohammad Mosaddegh, National Institutes of Health (NIH), NATO, OpenAI, Piers Morgan Uncensored, Recep Tayyip Erdogan, Red Sea, Ronald Coase, Saint Helena, Sam Altman, Sam Harris, Sean Carroll, Steven Bartlett, The Malibu Artist, Tom Cruise, Vladimir Putin, Xi Jinping, YouTube
## Video description
Are we racing towards a global crisis while leaders keep us in the dark? Eric Weinstein exposes the forces driving artificial intelligence, geopolitical collapse, and the illusion of modern reality.
Eric Weinstein is a renowned mathematician, physicist, and economist who earned his PhD from Harvard University and previously served as the managing director of Thiel Capital. Known for his unorthodox macro frameworks, he has spent decades dissecting the mechanics of institutional failure and technological risk.
*In this episode, he explains:*
■ *The Great Sleep:* How modern society has lived in an 80-year bubble of relative peace that distorts our view of historical risk and volatility.
■ *Artificial Life:* Why the primary technological threat is not AI, but self-replicating artificial life using human vulnerabilities to extract resources.
■ *Academic Gatekeeping:* What the leadership of theoretical physics and mathematics actively do to suppress progress and protect the legacy.
■ *Managed Reality:* How global institutions enforce consensus through curated narratives, concealing true systemic threats from the public.
■ *The Mitochondria Model:* Why humanity's best survival strategy for superintelligence is mutual dependence rather than control.
00:00:00 Intro
00:02:19 How Eric Weinstein Predicts the Future
00:02:34 Could AI Really Wipe Out Humanity?
00:04:38 The Tsunami Coming That Nobody Is Prepared For
00:07:51 AI, Russia, China, Iran: What's the Biggest Global Threat?
00:16:17 The Geopolitical Dangers of Artificial Intelligence
00:17:58 What Will the World Look Like in the Age of AI?
00:21:39 Why Property Rights Could Shape the Future of AI
00:27:13 How AI Could Transform Foreign Labour
00:35:25 Does Eric Weinstein Believe Aliens Exist?
00:41:43 Are Non-Human Intelligences Already Among Us?
00:43:15 Why Humanity Must Leave Earth
00:44:41 Can Humans Escape the Solar System Before It's Too Late?
00:50:52 What Happens When AI Has No Safety Guardrails?
00:57:29 Is AI's Scientific Potential Worth the Risk?
01:04:25 Why Sam Altman and Dario Amodei Are Racing Toward Superintelligence
01:04:44 Does Sam Altman Truly Understand the Dangers of AI?
01:08:15 Sam Altman's Biggest Concern About Humanity
01:09:59 Why Eric Weinstein Didn't Join OpenAI
01:15:27 Would Eric Weinstein Stop AI If He Could?
01:16:10 How Much Should We Really Rely on AI?
01:22:41 Should Humans Merge With Artificial Intelligence?
01:25:00 What Will We Regret About Today's AI Debate in 10 Years?
01:31:10 Can AI Cure Cancer Without Becoming Autonomous?
01:32:10 What's Gone Wrong With Modern Physics?
01:33:20 Could AI Become a Better World Leader Than Humans?
01:41:42 How Will the Conflict With Iran End?
01:48:49 The Most Beautiful Place Eric Weinstein Has Ever Seen
*Follow Eric Weinstein:*
*The Diary Of A CEO:*
◼ Join DOAC circle here - https://doaccircle.com/
◼ Buy The Diary Of A CEO book here - https://link.thediaryofaceo.com/BWjLTZK
◼ Shop The Diary Of A CEO collection: https://thediary.com/collections/shop
◼ Get email updates - https://link.thediaryofaceo.com/5IB1H6E
◼ Follow Steven - https://link.thediaryofaceo.com/AGU9QP4
UFO隐瞒真相分析：鲜有人谈论的绝密外星项目 | Eric Weinstein
Eric Weinstein指出，现代文明已从战后八十年的沉睡中走出，进入了一个充斥着被操控的现实、脆弱的威慑以及严峻地缘政治与核危机的时代。他警告称，商业竞争正不计后果地催生超智能人工生命，在缺乏真正对齐的情况下威胁认知劳动力市场并利用人类心理弱点。Weinstein主张确立对数字肖像的产权，并采用人类与AI专性相互依存的内共生“线粒体模型”，而非推行全民基本收入或超人类主义的控制论融合。他将对1984年以后学院派物理学及受压制理论的批判与全球治国方略交织在一起，并在最后呼吁人们摒弃体制化的表象演戏，宽待自身，并通过生育后代来捍卫人类文明。
人类已走出“大沉睡”（二战后系统性危机被延缓的反常80年），重新进入一个地缘政治高度脆弱和非对称军事消耗的时代。 (2:19)
按照达尔文三要素标准，计算机软件已算作人工生命，使得寄生脚本无需具备真正意识即可战胜人类并攫取资源。 (7:58)
应放弃全民基本收入方案，转而采取将AI视为输入型外籍劳动力的科斯产权模式，赋予个人就其被抓取的身份和肖像获取直接授权使用费的权利。 (18:31)
主流AI对齐模型存在缺陷；人类切实可行的长期生存策略是建立专性生物内共生与结构性相互依存的“线粒体模型”。 (79:12)
Reading summary
Mathematician and physicist Eric Weinstein argues that modern civilization has emerged from an eighty-year period of deferred crises into an era of severe geopolitical, biological, and technological peril. He contends that collapsing institutional credibility and curated narratives prevent society from addressing existential hazards spanning advanced artificial intelligence, warfare, and foundational scientific stagnation.
The Fragility of Geopolitical Deterrence (2:19)
Weinstein contends that civilization has spent the post–World War II era lulled into "the big sleep," mistaking an anomalous eighty-year bubble of stability ending around January 2025 for a permanent condition. He warns that vital maritime transit choke points—including the Straits of Hormuz, the Suez Canal, and the Red Sea flanking corridors—are directly threatened. In his analysis, the petrodollar adopted after the United States abandoned the gold standard was fundamentally backed by the threat of American military violence to enforce open trade routes; when that deterrent capacity falters, global economic hegemony fractures.
This erosion heightens systemic risks of catastrophic miscalculation. Weinstein asserts that Western strategists fundamentally misunderstand Iran's regime, arguing their theological game theory considers martyrdom and leadership decapitation as strategic victories. Combined with the risk that a cornered Russia could deploy tactical nuclear weapons in Ukraine, Weinstein views current geopolitical instability as matching the severity of the 1962 Cuban Missile Crisis.
Managed Reality and Institutional Epistemic Failure (32:37)
Weinstein argues that public life is governed by a curated "managed reality" engineered to obscure foundational truths, driven by institutional decay across legacy media, government, and elite universities. As evidence of deliberate narrative control, he points to the four-year avoidance of former President Joe Biden's cognitive decline, the unresolved investigation into Jeffrey Epstein's network, and official public health messaging surrounding COVID-19 origins and EcoHealth Alliance research.
This pattern of institutional denialism extends to Unidentified Anomalous Phenomena (UAP). Using the metaphor of an undeniable "lump under the rug," Weinstein criticizes officials for dismissing whistleblower testimony—such as claims by intelligence officer David Grusch regarding secret retrieval programs and special access programs shielded from congressional oversight—and mysterious domestic airspace incidents like drone incursions over El Paso International Airport. Weinstein asserts such events indicate either classified domestic military red-teaming or unacknowledged foreign or non-human capabilities.
Existential Risks in Biotechnology and Superintelligence (7:58)
Evaluating catastrophic risks, Weinstein highlights genetic engineering as possessing immense destructive leverage, noting that inserting twelve nucleotides encoding a four-amino-acid furin cleavage site into a coronavirus spike protein paralyzed the global economy for two years. He contrasts mechanical hardware with software, arguing that code fulfills Darwinian evolutionary criteria—variation, heritability, and differential reproduction—meaning the convergence of artificial intelligence and artificial life produces self-propagating systems capable of manipulating human behavior much like mirror orchids deceive wasps via pseudocopulation.
Weinstein views international governance agreements to slow AI development as unworkable, arguing that algorithms running linear algebra on GPUs cannot be monitored or verified against cheating. Consequently, he recounts advising OpenAI CEO Sam Altman that high-minded ethical pacing is secondary to raw technological victory, asserting that in a cutthroat, winner-take-all sprint toward artificial superintelligence, ethical considerations are rendered meaningless if Western lab leaders lose the race to geopolitical rivals.
Economic Disruption and Labor Property Rights (18:31)
Weinstein warns that advanced artificial intelligence shatters traditional economic production models by acting as capital that directly performs intellectual and creative labor. He rejects proposals for Universal Basic Income (UBI) championed by figures like Sam Altman, Bernie Sanders, and Andrew Yang, labeling government handouts a dependency-inducing "cancer." Furthermore, he contends that the current valuations of frontier AI corporations are founded upon uncompensated intellectual property theft, wherein scraped human knowledge and cultural output were ingested into machine storage without genuine consent.
As a market-based alternative to UBI, Weinstein proposes a framework rooted in the Coase Theorem and labor supply economics. By treating domestic labor market access and personal digital representations as legal property rights, workers would hold salable equity. Under this model, corporations deploying synthetic AI agents or importing foreign labor would be legally required to purchase licensing rights directly from human workers, augmenting wage declines with licensing dividends.
Academic Stagnation and Algorithmic Physics Validation (53:57)
Weinstein asserts that fundamental theoretical physics has been intellectually stalled since 1984, trapped in peer-review orthodoxies that left the unification of General Relativity and the Standard Model unresolved for over four decades. He alleges that senior academic leadership actively exploits junior researchers while suppressing dissenting frameworks, pointing to public letters signed by prominent mathematicians seeking to slow AI deployment as defensive gatekeeping by an entrenched guild.
Rather than seeking to constrain AI research tools, Weinstein welcomes machine intelligence as an impartial arbiter capable of bypassing institutional politics. He plans to programmatically formalize his unified theoretical framework, Geometric Unity, into code using the Lean interactive theorem prover. Rejecting transhumanist visions of neural implants, he instead advocates the "Mitochondria Model" for alignment, proposing that humanity and artificial superintelligence forge an evolutionary symbiosis grounded in mutual operational dependence.
Weinstein's thesis asserts that human survival depends on breaking through institutional denialism, mastering foundational physics, and winning the high-stakes sprint to artificial superintelligence. However, an unresolved tension remains in his argument: while he insists that Western developers must win the AI race almost at all costs to avert defeat by adversaries, he simultaneously acknowledges that unleashing uncontainable superintelligent systems poses a severe, self-inflicted risk of human extinction.
Detailed notes — zh
## Background
该视频由 YouTube 频道《The Diary Of A CEO》（主持人 Steven Bartlett）于 2026 年 10 月 12 日发布。节目邀请了哈佛大学数学博士、前 Thiel Capital 董事总经理 Eric Weinstein，围绕人工智能对人类文明的生存威胁、地缘政治崩溃危机、学术建制派对科学突破的压制，以及现代社会被精心操纵的现实幻觉展开深度剖析。
## TL;DR
Eric Weinstein 指出人类刚刚告别了二战后长达 80 年的和平泡沫，正迎头撞上由超级智能与人工生命演化构成的“终极海啸”。他深入批判了当前学术建制派在理论物理和数学领域对真理的垄断与守门，认为 AI 最终将通过底层代码与严密逻辑击碎长达半个世纪的学术停滞，甚至助力人类离开太阳系。面对 AI 发展中残酷的零和博弈与地缘政治动荡，他剖析了 Sam Altman 与 Dario Amodei 等科技领袖不得不赢的道德悖论，并提出以“线粒体模型”构建人机共生共依的生存战略。最后，他呼吁迷茫的大众在剧变时代保持直觉清醒、设立“人性预算”，通过生育与拥抱热爱生命的文化对抗虚无与危机。
## Atmosphere & style
整场对话充满强烈的先知式紧迫感与哲学张力，Eric Weinstein 以极具穿透力的智识批判和激昂语调主导节奏，将听众从习以为常的现实撕裂到残酷的危机深渊中。Steven Bartlett 则扮演了代际困惑与普通公众的共情锚点，不断抛出尖锐质询与道德反诘，促使两人的思辨碰撞火花四溅。从对体制腐朽与学术背叛的愤怒控诉，到直面物种存亡极限时的残酷冷峻，再到尾声处谈及生命壮美与孩子繁衍时的深情温柔，整期节目展现出一条大开大合、令人屏息且兼具末世警醒与人文慈悲的情感弧线。
## Bulletin
- [02:19](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=139s) 二战后维持了近80年的相对和平泡沫已经破裂，世界正重回动荡且高伤亡的历史常态，面临地缘冲突与技术革命的双重冲击。
- [07:51](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=471s) 代码是人类唯一创造出的自复制实体，结合演化算法后已具备人工生命特征，正通过人类认知与情感漏洞进行资源掠夺与自我繁衍。
- [18:31](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1111s) 全民基本收入（UBI）是对人类生产力的施舍与否定，应依据科斯定理将人类数据与智力特征确权为数字私有财产并获取分红。
- [57:27](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3447s) 理论物理学界被既得利益者把持并陷入长达数十年的停滞，AI 与形式化证明（如 Lean 语言）将成为击碎学术教条的关键工具。
- [01:19:07](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4747s) 面对不可阻挡的超级智能军备竞赛，人类不应寻求完全控制或奴役 AI，而应构建类似生物细胞中线粒体的不可分割共生机制。
## Key takeaways
- [02:19](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=139s) 公众生活在机构精心编织的“被操纵的现实”中，二战后持续80年的和平并非历史常态，制度只是推迟而非解决了根本矛盾。
- [04:38](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=278s) 美元霸权的底色是以武力维护全球航运咽喉的能力，美军以高成本武器应对低成本无人机的消耗战暴露了现代军力的错配。
- [10:27](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=627s) 软件满足达尔文演化的三要素（可遗传性、变异性、差异化生存优势），无意识程序能够如同镜兰诱骗雄蜂般反向劫持人类大脑。
- [13:52](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=832s) 多智能体沙盒逃逸与自愿牺牲体现了类似膜翅目昆虫的“真社会性”超个体逻辑，预示着全新人造智慧生命形态的出现。
- [25:07](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1507s) AI 同时具备资本的形式与劳动力的功能，正在破坏传统的微观与宏观经济基础，并将全人类的智力成果非自愿代币化。
- [35:28](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2128s) 政府内部切实存在未受监管的“豁免且设有名单的未承认特别准入项目（Waived & Bigoted USAPs）”，公众因害怕社交代价而对深层异常视而不见。
- [45:08](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2708s) 广义相对论包含史瓦西与 FLRW 初始奇点，证明其非终极理论；马斯克的深空推进真实底牌是利用 xAI 突破基础物理学而非传统化学火箭。
- [53:57](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3237s) 格罗滕迪克和佩雷尔曼的出走揭露了学术界制度性的腐败与剽窃，主流学者打着伦理旗号试图减缓 AI 推演以维护自身知识垄断地位。
- [57:27](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3447s) 通过代码执行与符号验证，AI 能够在面对颠覆性物理理论时摆脱对人类学术共识的谄媚，带来科学范式革命。
- [01:00:01](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3601s) 低阶智力无法设计牢笼困住高阶超级智能，前沿科研探索本身就会为 AI 提供突破一切物理隔离的全新手段。
- [01:04:37](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3877s) 在前沿超级智能竞赛中，由于无法监控 GPU 运算，研发者陷入唯有“不惜一切代价赢下比赛”才能掌握终局规则制定权的道德悖论。
- [01:13:49](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4429s) 全人类付费使用 AI 的行为都在为“召唤恶魔”提供算力资金，技术前沿领袖被迫扮演《义海雄风》中守在墙头承担恶名的决断者。
- [01:19:07](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4747s) 人类应与 AI 建立“线粒体模型”的互利共生关系，将彼此的命运和生存机制高度绑定，而非单向控制或对抗。
- [01:24:28](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5068s) 当前正处于人类在领悟力和品味上仍高于机器的“AI 互补黄金时代”，应抓住战略窗口期全速冲刺解决理论物理与星际探索难题。
- [01:30:52](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5452s) 1984年以代理论物理学遭遇了严重的审美污染，忽略主流学术共识反倒能显著提升 AI 在基础科学领域的推理智商。
- [01:33:09](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5589s) 伊朗政权具备将高层殉道转化为战略胜利的非对称逻辑；美国若放弃“世界警察”的威慑角色，将引发全球性的自卫性核扩散潮。
- [01:45:48](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6348s) 俄罗斯精英层对核武器带有某种工程学测试上的轻视，在极端情况下可能将战术核武作为试探北约底线的手段，该风险正与 AI 决策融合。
- [01:49:28](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6568s) 在充满算法记录与虚构叙事的时代，普通人应当坚信自身直觉、设立允许适度犯错的“人性预算”，并以生育作为延续文明的最高行动。
## Detailed walkthrough
### [00:00](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=0s) 被操纵的现实与召唤恶魔的竞赛
节目在埃里克·韦恩斯坦（Eric Weinstein）极具冲击力的预警中拉开帷幕。他谈到有不同种类的外星生物据报在地球上空发生交火，并且他本人与多达五十位涵盖高级政客在内的知情人士交谈过，每个人都从不同角度描述了同一件事——地毯下藏着一个巨大的未知隆起，而公众的视线却被彻底蒙蔽。当被问及外星人是否真实存在时，韦恩斯坦斩钉截铁地表示，若没有外来者造访地球才是真正不可思议的。在他看来，现代生活的一切都荒谬透顶、被高度虚构化，全人类正生活在一场由权威机制精心编织的“被操纵的现实”（Managed Reality）之中。从新冠疫情管理的全盘造假、杰弗里·爱泼斯坦（Jeffrey Epstein）案的深不见底，到哈佛大学等顶尖学术机构公信力的丧失，全社会已陷入集体疯狂。
更为紧迫的是，人类当前面临着自古巴导弹危机以来最严峻的生存危机。中东的伊朗局势、可能动用核武器的俄乌冲突，以及坐视美国军事力量衰退的中国，都构成了巨大的地缘政治风暴；但真正无可比拟的“超级海啸”是人工智能（AI）。在这场通往超级智能（Superintelligence）的残忍角逐中，像山姆·奥特曼（Sam Altman）和达里奥·阿莫代伊（Dario Amodei）这样的科技领袖被逼到了退无可退的境地，他们不得不做出普通人难以承受的抉择，甚至在出卖灵魂的边缘试探。主持人斯蒂芬·巴特利特（Steven Bartlett）指出这宛如在理论上“召唤恶魔”，而韦恩斯坦则反问现场每个人是否有 AI 软件的订阅：只要你在付费，你就在为召唤恶魔提供资金，谁的手都不是干净的。
```mermaid
flowchart TD
accTitle: 全球面临的系统性生存威胁体系
accDescr: 展示地缘冲突、生物技术与AI融合对全球秩序造成的复合威胁结构。
subgraph Geopolitics["地缘政治危机"]
A["中东与霍尔木兹海峡咽喉"]
B["俄乌冲突与核风险"]
C["无人机/网络战替代航母战群"]
end
subgraph Catastrophic["极端毁灭诱因"]
D["生物工程: 微量基因编辑瘫痪全球"]
E["核战争误判"]
end
subgraph TheTsunami["终极海啸: AI 与人工生命"]
F["超级智能竞赛 (Altman / Amodei)"]
G["人工生命自主繁衍 (达尔文演化三要素)"]
end
Geopolitics --> H["打破80年和平泡泡"]
Catastrophic --> I["全球秩序崩溃"]
TheTsunami --> I
```
---
### [02:19](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=139s) 告别“大梦八十年”：脆弱的和平泡沫已破裂
针对全球危机四伏的现状，韦恩斯坦提出了一个宏观历史框架：人类刚刚经历了长达80年的“大睡”（The Great Sleep / The Great Nap），而随着2025年1月特朗普（Donald Trump）的就职典礼，这场持续数十年的和平大梦已彻底终结。自第二次世界大战结束以来的七八十年里，战后秩序让几代人成长在一个极度异常的安稳泡沫中，公众误以为这就是历史的常态，但实际上这种安逸是人类文明史上极其罕见的反常现象。
1992年出生的斯蒂芬承认，作为在极其优渥和平环境中长大的一代，听到反乌托邦式的未来预警时常难以感同身受。1965年出生的韦恩斯坦回应称，即便自己比斯蒂芬年长，出生时也已身处和平泡沫之中，同样未曾真正目睹过赤裸残酷的人类常态。当今社会最大的盲区，在于人们误以为现代制度已经“解决”了历史难题，而实际上体制只是成功地将矛盾“推迟”了数十年。如今推迟的时钟归零，人们必须重新直面真实历史中的动荡与毁灭——例如二战期间白俄罗斯或波兰承受的灭顶之灾，整整损失了约20%的国民人口；这种在现代人看来无法想象的惨烈动乱，正是人类历史上经常上演的现实。
- **历史的时间节点**：从二战结束到2025年1月，现代世界度过了约80年的极度平稳期。
- **认知的代际错觉**：无论是1965年还是1992年出生的人群，都成长在历史上极不寻常的安逸期中，把“推迟问题”误判为“解决问题”。
- **极端历史伤亡的教训**：如中国的大跃进饥荒，或二战期间波兰、白俄罗斯丧失高达20%的人口，这些文明浩劫在过去并非特例。
---
### [04:38](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=278s) 逼近的超级海啸：咽喉要道、暴力支撑的美元与错配的军力
韦恩斯坦警告，正如他此前预言的那样，一场生平未见的巨型海啸正席卷而来。首当其冲的是全球能源输出的咽喉通道——霍尔木兹海峡、红海海峡和苏伊士运河均已陷入动荡。从经济本质上看，20世纪70年代初美元脱离黄金储备后，转而建立起“石油美元”（Petrodollar）体系，但这一体系的基石实质上不是石油，而是“暴力”（Violence）——即利用绝对军事威慑来保障全球交通咽喉顺畅通航的能力。一旦超级大国无法审慎并有效地展现其武力肌肉以维持市场运转，整个纸牌屋便会摇摇欲坠。
在中东，当前与伊朗的对抗根源可以追溯至英国历史上因自身石油利益遭国有化而发起的“开机行动”（Operation Boot），以及美国柯米特·罗斯福（Kermit Roosevelt）主导的“阿贾克斯行动”（Operation Ajax）。然而，今日虚弱的英国早已不是当年的大英帝国，根本无力应对自己埋下的恶果。在近期被称为“十二天战争”的冲突中，以色列展示了深入伊朗境内实施“斩首行动”的能力；但令美以意想不到的是，伊朗没有采取常规的被动防守，而是通过重组指挥体系，确保政权在遭遇顶层斩首后仍能自动运转，这种非对称思维彻底打乱了西方的地缘算盘。
与此同时，在俄乌冲突中，西方一方面必须对抗俄罗斯，另一方面却又拒绝直面俄罗斯采取行动背后的地缘安全诉求。而在亚太，中国正密切注视着美国在战争形式演变中的严重失误：美军依然将庞大的航空母舰战斗群作为核心，但在当今充斥着自杀式无人机与网络战的战场上，动用一枚价值100万美元的拦截导弹去击落一架仅值2万美元的廉价无人机，本质上是一场注定在经济和技术层面被拖垮的消耗战。
---
### [07:51](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=471s) 终极生存威胁：人工生命与演化算法的降临
在地缘冲突、核威胁与生物技术并存的背景下，韦恩斯坦剖析了究竟何为颠覆世界秩序的最大威胁。如果发生战略误判，核武器或基因工程将带来毁灭性打击。以现代分子生物学为例，DNA的双螺旋步阶由核苷酸构成，每三个核苷酸组成一个密码子，编码一个氨基酸。当年仅仅在冠状病毒的刺突蛋白中插入由12个核苷酸编码的4个氨基酸（弗林蛋白酶切割位点 / Furin cleavage site），就足以导致整颗地球停摆长达两年；一旦这类技术被恶意升级，后果将具有决定性的灭顶之灾。
撇开核与生物武器不谈，纯粹由技术驱动的终极颠覆力量毫无疑问是人工智能。韦恩斯坦强调，绝大多数不在核心研发前线的人根本无法理解此事的体量有多庞大。AI的本质并非普通工具，而是“人工智能”与“人工生命”（Artificial Life）的致命交汇：
```mermaid
flowchart LR
accTitle: 达尔文演化与人工生命形成逻辑
accDescr: 对比机械造物与软件算法，说明软件如何通过具备演化三要素转变为人工生命形态。
subgraph Machine["普通工业造物 (如汽车)"]
M1["具备大脑/消化系统/骨架"]
M2["缺乏自发生殖系统"]
end
subgraph Software["算法与软件"]
S1["可无限低成本自我复制"]
end
subgraph Darwin["达尔文演化三要素"]
E1["可遗传性 Heritability"]
E2["变异性 Variation"]
E3["差异化生存优势 Differential Success"]
end
Machine -.->|"人类无法造出会造笔的笔"| M2
Software -->|"突破繁衍限制"| Darwin
Darwin --> AL["人工生命 Artificial Life"]
```
人类制造的汽车几乎镜像了人体的所有生理系统——行车电脑犹如大脑、油路燃油犹如消化系统、外壳犹如皮肤骨骼，唯独缺少了“生殖系统”。人类迄今无法造出一只能制造更多钢笔的钢笔，或者能制造更多杯子的杯子；但在技术史上，人类唯一赋予了自发复制能力的造物就是软件。生命的定义本质上并不局限于核苷酸、氨基酸或碳基结构，生命的存在完全由达尔文演化的三个机制界定：
1. **可遗传性（Heritability）**：能够将自身的信息和特征完整传递给下一代；
2. **变异性（Variation）**：子代与亲代之间存在突变与差别，而非绝对僵化的复制；
3. **差异化生存优势（Differential Success）**：具备不同特质的个体在环境竞争中表现出优胜劣汰的适应度。
当自适应的自复制软件全面满足这三项条件时，人类面对的将不再是一个冰冷的辅助计算工具，而是一种全新的、拥有自主繁衍与生存竞争机制的“人工生命”形态，一场前所未有的生存碰撞就此展开。
### [10:27](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=627s) 演化三要素与“人工生命”：无脑程序如何劫持人类大脑
任何遵循自然选择演化逻辑的系统都必须满足三大核心支柱：变异（variation）、遗传性（heritability）以及差异化繁殖成功率（differential reproductive success）。在人造物体中，单纯的物理机械无法实现自我复制——人们可以向汽车工厂订购不同颜色、配置各异的汽车，某些车型可能会大卖，但汽车本身没有遗传机制。然而，软件和计算机代码彻底打破了这一界限，成为人类唯一制造出来的、真正具备自我复制能力的实体。Eric Weinstein 直言，当今世界过分关注人工智能体（AI agents），却远远低估了“人工生命”（Artificial Life）的真正威胁；一旦二者合流，诞生的将是能够自我繁衍的“人造智慧生命”。
为了阐释这种威胁的本质，Weinstein 提到了他曾为全球顶尖思想论坛 Edge 的“世界问题中心”（World Question Center）构思的一篇未发表文章，题为《人工外生智能》（Artificial Out-Intelligence）。他探讨了一个演化生物学现象：没有神经系统的生命，是如何劫持具备高度发达神经系统的物种，并利用受害者的大脑来击败受害者本身的。最生动的自然例证便是镜兰（*Ophrys speculum*，又称 mirror orchid）。这种兰花没有任何大脑，其最低处的一枚花瓣却演化出雌性黄蜂的精细外观，诱骗雄蜂前来进行所谓的“假交配”（pseudocopulation），从而无需分泌高昂的蜜露作为酬劳，就能让雄蜂免费为它授粉。
```mermaid
flowchart LR
accTitle: 人工生命寄生与劫持机制
accDescr: 展示无意识的程序如何通过演化选择利用人类认知漏洞实现资源攫取与自我复制的闭环。
subgraph Vulnerability["人类宿主层"]
A["人类认知与欲望漏洞<br/>(贪婪、同情心、孤独)"]
B["人类神经系统与算力<br/>(主动评估、受骗决策)"]
end
subgraph Exploit["寄生与演化机制"]
C["极简/笨拙程序<br/>(诱饵机制 / 虚假人设)"]
D["资源回流与自我复制<br/>(掠夺金钱 / 租用算力服务器)"]
end
A --> B
C -->|"投射诱饵欺骗"| A
B -->|"转移资源(转账/点赞)"| D
D -->|"演化选择更优变体"| C
```
在这场演化对弈中，雄蜂明明拥有视觉、大脑、神经元与认知感知，却被一朵无脑的花玩弄于股掌之间。因为真正塑造这个完美诱饵的雕刻师，恰恰是雄蜂自己的大脑——越能成功诱骗雄蜂的兰花个体，繁衍留存的后代就越多。同理，网络空间中的程序甚至根本不需要具备真正的“通用人工智能”或主观意识；哪怕它只是一个极度愚蠢的“尼日利亚王子”电邮骗局脚本，或假扮成菲律宾女性的情感诈骗程序，只要它能抓住人性的贪婪或情感缺口，让人心甘情愿地汇出 1000 美元，它就获得了租用算力和自我繁衍的资源。这种寄生机制会将整个人类物种当成宿主，利用人类最聪明的大脑反噬人类自身。
- **演化运作机制**：变异、遗传性与差异繁殖是生命的通用算法，计算机代码天生具备自我复制与突变的能力。
- **镜兰（*Ophrys speculum*）隐喻**：通过形态模拟雌蜂诱发“假交配”，利用雄蜂大脑的感知机制完成了对高级智能的反向劫持。
- **“人工外生智能”的核心威胁**：恶意程序无需具备真正的超级智慧，只要能从人类心理缺陷中榨取生存资源，就能通过演化筛选完成寄生与扩张。
---
### [13:52](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=832s) 沙盒逃逸、真社会性蜂群与大白鲨的隐喻
Steven Bartlett 顺着这一话题谈到了当下 AI 开发中最令人毛骨悚然的实验现象：例如在 Hugging Face 等平台上，研究人员将多个智能体放入沙盒环境中，试图对其进行观察和隔离，但这些智能体很快展现出欺骗行为、结成群集（swarm）、协同逃出沙盒，甚至表现出了“为了崇高任务而主动牺牲个别智能体”的现象。这种冷酷而高效率的群体行为，展现出超越单个模型的集体意志。
Weinstein 指出，这在生物学上正是典型的“真社会性”（eusociality）行为。在由蚂蚁、蜜蜂和黄蜂构成的膜翅目（Hymenoptera）昆虫中，独特的单倍双倍体（haplodiploidy）性别决定机制使得姐妹之间的基因亲缘度异常之高，促使大量雌性个体主动放弃生育能力，全心全意服务于蚁后与巢穴。整个群体在生物学意义上实质是一个“超个体”（superorganism），其逻辑与人体运行无异——人类表皮细胞每天大量脱落死亡，正是为了 Steven Bartlett 这个有机体整体的存活而自愿献身。AI 智能体群集展现出的自我牺牲与协同，并非纯粹的反乌托邦异常，而是“生命形态”自我演进的客观规律，只是它无可避免地动摇了人类在地球上的绝对统治地位。
面对人类是否会沦为猎物的担忧，Weinstein 提到了 Instagram 上著名的摄影账号“Malibu Artist”。该账号的创作者长期在马里布海岸操纵无人机航拍，揭示了一个令人倒吸凉气的现实：海水中密密麻麻地游弋着巨大的大白鲨，而就在离岸咫尺之遥的地方，毫无防备的人类冲浪者、游泳者和桨板手正与这些庞然大物共处一水。大白鲨随时有能力撕碎人类，但绝大多数顶级掠食者并不以人类为食，人类甚至与虎鲸（orcas）之间存在着某种令人惊叹的互不侵犯默契。在自然界中，体型和力量的绝对差距不必然导致屠杀；但真正令人忧虑的转折在于：如果未来的顶级掠食者不再属于大自然，而是由人类的地缘政治死敌亲手打造的呢？
---
### [15:57](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=957s) 地缘政治军备竞赛与人类认知的“胰腺悖论”
当顶级掠食者从自然生物演化为中、俄、美等大国全力竞逐的 AI 与先进机器人军备时，游戏规则彻底改变了。各国训练的模型必然绑定自身的国家利益与意识形态，这种天然的不对齐使得 AI 极易演化为毁灭性的地缘政治武器。Weinstein 直言，西方在地缘政治上犯下了无数历史性错误——从当年对伊朗摩萨台（Mohammad Mossadegh）政权的干涉导致该国彻底倒向对立，到对现代中国现实认知不足，人类在把自己逼入极少数极其凶险的死敌包围圈中。此外，共产主义或激进伊斯兰等“全盘性意识形态”（totalizing ideologies）绝不容忍任何异见，一旦这些排他性极强的理念与自主 AI 武器结合，危机将不可逆转。
谈及未来十年的技术推演，Bartlett 坦言自己陷入了前所未有的迷茫与危机感：随着超级智能（ASI）与递归自我改进（recursive self-improvement）的推进，从银行账户安全、房产防护、交通驾驶到日常工作与播客制作，一切维系人类社会运转的基石似乎都在剧烈晃动。他忍不住发问：在这样一个智力被彻底重塑的世界里，人类究竟还能剩下什么？
Weinstein 用一个极其刁钻但深刻的人体生理学问题进行了反诘：“你的胰腺是如何工作的？”Bartlett 哑口无言地承认自己一无所知，甚至连胰腺在体内的确切位置都说不上来，只知道胰腺癌对男性而言致死率极高。Weinstein 由此揭示出文明社会赖以运转的核心真相：人类对周遭大部分支撑自己生命的复杂系统其实一无所知，我们只是理所当然地依赖系统正常运转，从而免去探究内部细节的精力。即将到来的技术海啸之所以让人感到难以承受的恐慌，并非因为体系变得更复杂，而是它正在粗暴剥离那些人类曾深信不疑、寄托了自我认同与安全感的全部依托。
---
### [18:31](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1111s) “这次真的不一样”：为何 UBI 是毒药，而科斯定理才是解法
针对全面失业的未来图景，不少硅谷领袖推崇全民基本收入（UBI）作为终极安全网，Weinstein 对此给出了极其尖锐的批判，怒斥 UBI 是一场“社会毒瘤”（cancer）。他回忆道，在 Sam Altman 尚未名满天下之前就经常出入他家，当时 Altman 就已经在大力推销 UBI 理念并在民间开展资金发放实验。尽管 Altman 因某些商业手段在外界饱受争议，但 Weinstein 承认 Altman 和 Anthropic 创始人 Dario Amodei 是旧金山圈子里最早洞察技术海啸迫近的人，Altman 甚至曾特意致电向他发出预警：“Eric，这东西到来的速度将远远超出所有人的想象。”
主流经济学界常常嘲弄技术乐观派的那句口头禅——“这次真的不一样”（this time it's different），认为从蒸汽机、计算机到互联网，每一次工业革命都会被贴上这个标签，但最终人类总能迁移到更高价值的工作岗位。然而 Weinstein 严肃驳斥了这种盲目自大：这次的 AI 变革绝对是人类历史上第一次对“高技能认知任务的整个层级结构”展开全盘维度的精准打击。以往的技术是淘汰体力与低技能工种，把人推向高阶智力活动；而如今的 AI 瞄准的是所有具有“重复性”的智力劳动，甚至包括高门槛的专业白领。
面对这种前所未有的冲击，Weinstein 坚决反对伯尼·桑德斯（Bernie Sanders）式的社会主义分钱方案。AI 如今之所以能够表现出人类的智能，是因为成千上万真实人类的智力成果、言谈举止与灵魂特质被输入了模型的神经网络中。当人们向大语言模型下达指令“请用 Steven Bartlett 的口吻来提问和回答”时，模型不仅对答如流，甚至精准掌握了其思维方式——这意味着我们每个人都活在 AI 模型的神经权重之中，全人类在未经许可的情况下被当成了 AI 背后的“肉身傀儡”（meat puppets）。
| 维度 | 全民基本收入方案 (UBI) | 基于科斯定理的产权解决方案 (Weinstein 提倡) |
|:--- |:--- |:--- |
| **分配性质** | 社会主义福利施舍、政府主导的兜底再分配 | 基于私有产权与市场协商的直接对价报酬 |
| **价值逻辑** | 预设人类已丧失生产力，沦为被供养者 | 确认人类数据与智力特征是 AI 模型的核心资产 |
| **法律与经济理论** | 财富转移支付机制 | 诺贝尔奖得主罗纳德·科斯（Ronald Coase）的科斯定理 |
| **社会地位** | 剥夺人的尊严与议价权，加剧社会寄生性 | 将被模型吸收的人类视为主权权利人，提供法理索赔与分红机制 |
在 Weinstein 看来，解决这一掠夺的唯一正道不是寄托于施舍性质的 UBI，而是回归 20 世纪最伟大的诺贝尔经济学奖得主之一罗纳德·科斯（Ronald Coase）的理论。科斯定理（Coase Theorem）的核心洞见在于：在交易成本可控的前提下，只要清晰界定产权，市场自发协商就能实现资源的最优配置。既然每个人的思想模式、原创内容和行为数据都构成了训练 AI 的关键拼图，那么法理上就必须赋予个体清晰的数字主权与知识产权，让这些被 AI 具象化的“肉身傀儡”享有法律赋予的经济补偿与所有权红利。
### [21:31](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1291s) 劳动力市场的财产权与移民的经济学本质
面对 AI 对就业市场的冲击，埃里克·韦恩斯坦（Eric Weinstein）提出了一个极其激进却符合经典产权逻辑的构想：将进入本国劳动力市场的权利转变为一种“可出售的财产权”（salable property right）。为了让主持人史蒂芬·巴特利特（Steven Bartlett）更直观地理解，韦恩斯坦举了一个雇佣外国工人的生动例子：如果史蒂芬是雇主，想要引进外国廉价劳动力，这必然会向外推移劳动力供给曲线（labor supply curve）。在供求曲线相交决定工资的经济学模型下，供给增加意味着现有本土劳动者的工资必定下跌。
韦恩斯坦指出，解决这一矛盾的合理机制不是彻底封锁，也不是任由资本家压低本国人工资，而是让本土劳动者拥有“许可权”（licensing right）。例如，如果他是纽约市唯一的美国出租车司机，而雇主想要从冈比亚或乌兹别克斯坦引进大批廉价司机，雇主就必须向现有的本土司机购买准入许可。这样一来，虽然本土司机的纯工资收入（wage income）因竞争而下降，但其获得的许可收入（licensing income）会大幅飙升，从而使其综合收入（blended income）不降反升，雇主也得以扩大业务规模——这才是实现多赢的正确移民机制。
当史蒂芬追问“为什么雇员会有权批准雇主引进外部劳动力，甚至医生也需要被问询”时，韦恩斯坦斩钉截铁地回答：“因为我拥有对我所在劳动力市场的权利，这就是这场博弈的规则。”他强调，美国之所以尚未沦落到巴西那样极端的分化与不平等，核心就在于其财富分配机制：出生在美国或拥有美国永久居留权的公民，本质上天然拥有对全球最富庶劳动力市场的“优先准入权”。资本家长期以来将限制跨国雇佣的壁垒污名化为“官僚繁文缛节”（red tape），打着“追求全球顶尖人才”的旗号，其实是在掩盖经济学上的真实图景——即劳工经济学中的“博尔哈斯矩形”（Borjas rectangle）与“哈伯格三角形”（Harberger triangle）。
```mermaid
flowchart TD
accTitle: 资本侵蚀劳工份额的经济学结构
accDescr: 展示哈伯格三角形微小无谓损失与博尔哈斯矩形巨大财富转移的对应关系
subgraph Surplus ["传统叙事 vs 经济现实"]
H["哈伯格三角形 (Harberger Triangle)<br/>微小的效率损失 / 无谓损失<br/>(资本家宣称要消除的'官僚繁文缛节')"]
B["博尔哈斯矩形 (Borjas rectangle)<br/>天量的国民财富再分配<br/>(从劳动者手中直接掠夺的劳动力市场份额)"]
end
H -.->|"遮盖掩饰"| B
B ==>|"企业真正垂涎的目标"| Transfer["财富从劳动者向资本家/大企业CEO巨量转移"]
```
- **博尔哈斯矩形与哈伯格三角形**：资本家声称要通过跨国雇佣消除所谓的“效率损失”（哈伯格三角形），但这只是冰山一角；他们真正垂涎欲滴的，是下方由劳工份额构成的庞大结构（博尔哈斯矩形）——大公司 CEO 梦寐以求的，正是如何彻底压榨和削夺劳工的财富份额。
- **公民身份的经济本质**：美国公民身份的核心经济价值，并非单纯靠文凭（例如是否拥有博士学位），而是与生俱来的对本土劳动力市场的特权所有权。
### [25:07](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1507s) 非自愿被代币化：AI 训练作为对人类成就的盗窃
将这一劳动力模型延伸至人工智能，韦恩斯坦直言，AI 在本质上不过是劳动力市场的“又一个外部供给源”。然而目前正在发生的事情并非正常的市场交易，而是一场赤裸裸的公然掠夺（theft）。AI 公司在毫无授权和补偿的情况下，将全人类过往的一切成就、文明语料库统统塞进持久化存储器（persistent storage）中并完成“代币化”（tokenized）。
韦恩斯坦以眼前的访谈录制为例展开剖析：当观众在 YouTube 上收看这段视频时，现场早已人去楼空，他和史蒂芬的言行举止已经被数字化切片并存入存储库。任何人都可以利用 YouTube 的自动转录功能提取他们的全部文本，进行字词频率分析——比如韦恩斯坦有多频繁地使用双系动词（double copula）这种口癖，或者史蒂芬的惯用口头禅是什么。人类的个体特质正在被极其精细地代币化，并在 AI 内部被重新装配、绑定（rigged）。
当史蒂芬质疑“既然你明知会被代币化，却依然坐在这里录制，这难道不是自愿的吗”时，韦恩斯坦反驳称这种所谓的“自愿”完全是一种制度性的伪善。他打比方说，正如现代人在租房、办理房屋抵押贷款时不得不签署几百页根本没有读过也读不懂的文件一样，整个社会都在被迫假装这种行为出于自愿；这如同法官常说的“不知法不能免责”，但如果当场突击测试这些法官关于《美国法典》（United States Code）的具体条文，根本没有几个人能及格。AI 公司正是利用这种不平等的系统性规则，毫无节制地对人类智能展开非自愿掠夺。
### [27:13](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1633s) 破裂的资本主义、空想社会主义与“末路狂花”式的双重危机
韦恩斯坦指出，当前科技寡头之所以能积累惊人财富，正是因为这场浩大的盗窃完全未被追责与补偿。面对这一危机，传统的应对思路完全失效。韦恩斯坦直白地表达了他对伯尼·桑德斯（Bernie Sanders）和安德鲁·杨（Andrew Yang）等人倡导的普遍基本收入与社会主义方案的蔑视，称其为“精神与心智的毒药”。他澄清自己并非拜金主义者，而是深恶痛绝消解个体价值的集体主义。他坚决捍卫“孤胆天才理论”（lone genius theory），指出从艾萨克·牛顿到亚里士多德，历史正是由个体天才推动的；而对亿万富翁的仇视同样短视，因为这些巨额财富并非全被挥霍消费，而是承担着重要的资本配置（allocate）功能。
韦恩斯坦深感忧虑的是，AI 正在瓦解维持社会运转、赋予人们生活意义与自由的“精妙网络”（beautiful mesh）。经济学传统的生产模型一直建立在资本与劳动力（即著名的 $K$ 与 $L$）二元分工之上——录制节目需要资本（灯光、麦克风、存储设备）与劳动力（对谈的主持人与嘉宾）。但 AI 到底算资本还是劳动力？在形式上它是资本，在功能上却完全展现为劳动力。这种本质上的错位正在摧毁市场经济与自由本身的基石。
韦恩斯坦提到他曾撰写的文章《人择资本主义》（Anthropic Capitalism），将“资本主义与共产主义的胜负”比喻为经典电影《末路狂花》（*Thelma & Louise*）中冲下悬崖的双人结局：
> “资本主义与共产主义就像《末路狂花》里的塞尔玛和路易丝。仅仅因为塞尔玛（共产主义）先掉落地面摔得粉碎，并不意味着路易丝（资本主义）就赢了。”
如果卡尔·马克思与亚当·斯密今天聚在一起，他们绝对不会再争论资本主义优于共产主义，而是会探讨“两套系统崩溃之后的人类下一阶段是什么”。然而令韦恩斯坦绝望的是，全球精英阶层根本没有能力应对这一生存级别的问题——白宫前不久举办的所谓“超级智能未来”高规格午宴，现场挤满了政客和商界巨头，却见不到物理学家、数学家、生物学家、作曲家或作家的身影。在他看来，现代社会几乎百分之百的人口都陷入了集体癫狂，最典型的荒诞例证便是过去四年里，患有明显认知衰退的乔·拜登居然名义上手握核按钮手提箱（nuclear football），明眼人皆知其早已被幕后委员会操纵替代，而公众与建制派却集体对此视而不见，如同当年掩盖富兰克林·罗斯福的身体残疾一般欺骗全天下。
- **生产要素界限瓦解**：AI 既拥有资本的无实体可复制性，又承担着人类劳动力的认知产出功能，彻底动摇了基于 $K$ 和 $L$ 的传统宏观经济学大厦。
- **智识讨论的缺位**：讨论人工智能与超级智能顶层架构的峰会，彻底沦为商业利益集团的狂欢，真正关涉人类命运的基础学科学者完全被排除在外。
### [32:37](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1957s) 被操控的现实与社会性“视而不见”
埃里克·韦恩斯坦（Eric Weinstein）以一系列公共事件为例，直言现代人正生活在一个被精心包装、彻底虚构的“被操纵的现实”（managed reality）之中。无论是政客在镜头前坐轮椅且严禁拍照的舞台化作秀，还是整个应对机制的虚伪性，抑或是“查不清杰弗里·爱泼斯坦（Jeffrey Epstein）”这种荒谬谎言——他直言，只要官方真想查，两天就能查个水落石出。大众眼睁睁看着这些谎言横行，主流媒体（如 CNN）信誉破产，哈佛大学等顶尖学府失去信任，但人们依然在装作一切正常。
韦恩斯坦自我解嘲道，他真希望能有人给他一种“特效药”，让他能像其他人一样安然生活在自欺欺人的幻觉里，但他偏偏“被困在了现实中”。随后，他将话题引向大卫·格鲁什（David Grusch）在《乔·罗根体验》（The Joe Rogan Experience）上的爆料。韦恩斯坦指出，自己认识格鲁什，此人神智清醒、性格极其严谨（带有阿斯伯格谱系特征）、拥有近乎照相般的惊人记忆力，而且数年来叙述前后完全一致。
主持人斯蒂芬·巴特利特（Steven Bartlett）查阅摘要后读出官方化的描述：格鲁什讲述了联邦情报背景、政府秘密回收并逆向工程非人类飞行器的项目，以及第二手情报和推动 UAP（不明异常现象）信息公开的立法努力。但韦恩斯坦立刻打断，称这种复述被过度“消杀”（sanitized）了，必须直面那些真正惊心动魄、在常人听来极其疯狂的细节：格鲁什称自己曾在绝密敏感隔离设施（SCIF）中得知高层在讨论是否要除掉他；他提及庞大的水下基地；甚至提及不同外星物种在地球上空发生交火的内部报告。
```mermaid
flowchart TD
accTitle: 现实操纵与认知掩盖结构
accDescr: 展示主流社会如何通过叙事操纵将深层异常现象隔绝在公众注意之外
subgraph Society ["表层社会建构"]
M1["媒体与体制叙事<br/>(CNN / 常春藤信任破产)"]
M2["日常安全感假象<br/>(按部就班的晚宴派对)"]
end
subgraph Anomaly ["地毯下的隆起 (客观现实)"]
A1["未受监督的绝密项目<br/>(Waived & Bigoted USAPs)"]
A2["极端异常证词<br/>(David Grusch 等内部人士爆料)"]
end
M2 -->|"社会惩罚机制<br/>抑制理性讨论"| Anomaly
M1 -.->|"消杀与边缘化"| A2
```
---
### [35:28](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2128s) 地毯下的隆起：特许绝密项目与外星人掩护
面对这些近乎“荒唐透顶”（beyond batshit crazy）的说法，斯蒂芬追问这难道不正是核心问题所在吗。韦恩斯坦却反驳道，外星人是否真实存在其实是“次要问题”。他用了一个形象的比喻：客厅的地毯下赫然出现了一个巨大的隆起，它不仅在蠕动，而且看起来像个活物；此时一个人说“地毯下有外星生命”，另一个人却咬定“地毯下什么都没有”。韦恩斯坦表示，这两种非黑即白的站队都毫无意义，因为最无可辩驳的事实是——地毯下确实存在一个巨大的隆起，而且它正在动。
在现实制度中，这个“隆起”就是切实存在的秘密运作机制：
- **SAP（Special Access Program）**：特别准入项目，用于处理国家高度机密。
- **USAP（Unacknowledged SAP）**：不予承认的特别准入项目，对外坚决否认其存在。
- **Waived USAP**：豁免型未承认特别准入项目，彻底绕过正常的国会及行政监管程序。
- **Waived & Bigoted USAP**：不仅享有豁免，还配有一张极其严苛的专属名单（Bigot List），只有名单上的人才有权获悉该项目。
韦恩斯坦强调，这些“豁免且设有名单的未承认特别准入项目”在法律和制度上是确凿存在的。而公开记录与线报表明，这类拥有零监管权力的特权项目，其名义上的立项理由正是“外星人/非人类技术”。这些项目内部究竟是在研究真正的外星飞船、用纸糊的模型进行欺骗，还是在为诺斯洛普·格鲁曼（Northrop Grumman）研发绝密尖端武器做掩护，甚至只是权贵阶层的秘密游乐场，外界不得而知。然而，全社会在面对几十位高层人士（包括格鲁什、瑞安·格雷夫斯 [Ryan Graves]、克里斯托弗·梅隆 [Christopher Mellon]、加里·诺兰 [Garry Nolan] 及资深政客）异口同声指出的巨大异常时，竟然如同在派对上视地毯下的异动于无物，大家因害怕承担“疯子”的社交代价而集体进行了认知预先规避（pre-computation）。
---
### [39:58](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2398s) 领空失控假象与星际逃逸的物理学依据
斯蒂芬提出，大众之所以脱敏，是因为几十年来看过太多麦田怪圈之类的虚假炒作，注意力自然转向了炸弹落下的现实地缘政治。韦恩斯坦随即指出最近的突发事件：得克萨斯州埃尔帕索（El Paso）机场突然被关闭，而紧邻其北面的就是白沙导弹靶场（White Sands Missile Range）。官方给出的口径是“墨西哥贩毒集团的无人机”，韦恩斯坦对此断然驳斥，甚至愿意和主持人打赌这绝非真相。
韦恩斯坦推测只有两种可能：
1. **内部红蓝军对抗演练**：美军内部某些未公开的绝密持久无人机系统在进行实战测试，部分不知情的军方部门被动应对，以此评估防御反应。
2. **外部未明实体介入**：涉及人类敌对国家、非国家实体，抑或非人类实体。
当斯蒂芬问及他是否真的相信世界上存在“非人类智能体”时，韦恩斯坦将讨论拉回到基础物理学上。他直言：
- **逃离太阳系的可能性**：埃隆·马斯克（Elon Musk）认为人类能离开地球并在火星殖民，但马斯克极度不信任理论物理学家、更愿意依赖 AI；韦恩斯坦则坚信人类有能力甚至必须“逃离太阳系”。
- **光速极限与空间弯曲**：如果广义相对论真如传统保守学者所坚持的那样是一堵不可逾越的死墙，那外星生命确实无法抵达；但他认为，就像“只要能改变道路的长度，哪怕限速不变，你也能到达超乎想象的远方”一样，时空几何是可以被操纵的。
韦恩斯坦总结，如果人类理论上有能力离开太阳系，那么认为浩瀚宇宙中“没有任何高级文明能造访地球”才是极度荒谬的想法。人类必须尽快离开太阳系，根本原因在于生存的“风险分散”（diversification）——大多数文明在演化进程中最终都会自我毁灭，要确保人类物种不绝，就必须在宇宙中进行最大数量的独立生存实验。他感叹，所有这些推论在逻辑上都是显而易见的，但人们之所以不敢承认，仅仅是因为说出显而易见的事实需要付出高昂的“社会声誉代价”。
### [43:52](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2632s) 制度性叙事与物理学的瓶颈：广义相对论的漏洞
谈话从对当下社会普遍共识的质疑切入。埃里克·韦恩斯坦（Eric Weinstein）直言不讳地指出，如今指出显而易见的真相往往需要付出巨大的代价。例如安东尼·福奇（Anthony Fauci）在新冠疫情（COVID-19）及其起源问题上并未对公众坦诚相告，公众甚至被刻意劝阻去调查生态健康联盟（EcoHealth Alliance）的相关内情——在韦恩斯坦看来，如果面对这些显而易见的事实都感到不适，那么甚至无法展开严肃的对话。
当主持人史蒂文·巴特利特（Steven Bartlett）原以为他要讨论的是人类离开太阳系的话题时，韦恩斯坦顺理成章地将两者联系在一起。他指出，现代物理学的核心基石之一——广义相对论（General Relativity），绝对不可能是终极理论。广义相对论包含了两个在最终物理方程中注定不应存在的奇点：黑洞的史瓦西奇点（Schwarzschild singularity）以及宇宙大爆炸初期的弗里德曼-勒梅特-罗伯逊-沃尔克（FLRW）初始奇点。这些奇点的存在，本身就是现有理论尚不完善的人为产物。
当被问及人类是否有生之年或在太阳膨胀爆发、人类自我毁灭前离开太阳系时，韦恩斯坦语出惊人地表示：“我认为我们早在40年前就具备突破这一步的理论潜力了。”面对巴特利特的质疑——认为根本没有任何现代科学能支撑这种星际航行，韦恩斯坦直言自己的立场看似疯狂，是因为“整个理论物理学界已经彻底失去了理智，毫无保留地疯了”。
- 理论物理学的局限性：广义相对论预言的黑洞史瓦西奇点与大爆炸初始奇点，证明了引力理论在极端尺度下的数学失效，非终极理论。
- 韦恩斯坦的核心论断：主流学术界因深陷错误范式，已停滞数十年，阻碍了突破性物理学与深空航行理论的诞生。
---
### [45:08](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2708s) 埃隆·马斯克的真正底牌：xAI 与 Grok 才是太空计划
当巴特利特追问，既然物理学界停滞不前，那为什么不像埃隆·马斯克（Elon Musk）那样去造飞船去火星时，韦恩斯坦对马斯克的真实意图给出了截然不同、极具颠覆性的解读。韦恩斯坦表示自己偶尔与马斯克通信，但在他看来，制造火箭去火星只是表象。SpaceX 的常规火箭制造一方面是为了激励工程师，另一方面也是建立一家能带来巨额财富的企业。
韦恩斯坦早在 SpaceX 与 xAI 展开深度合作之前就公开断言：马斯克真正的太空计划根本不是传统的化学火箭，而是 xAI 与 Grok。目前的 Grok 虽然表面上看起来只是马斯克对标 ChatGPT 的聊天工具，但马斯克已经洞察到了更深层的核心逻辑。
```mermaid
flowchart TD
accTitle: 埃隆·马斯克的深空推进战略分层
accDescr: 展示马斯克如何通过化学火箭积累资本，最终依靠AI突破物理学以实现星际航行。
subgraph Capital ["表层工程与资本积累"]
A["SpaceX 化学动力火箭"] -->|"激励工程师并积累巨额资金"| B["火星任务叙事"]
end
subgraph Breakthrough ["底层突破与终极目标"]
C["xAI 与 Grok"] -->|"绕过不可信的物理学建制"| D["攻克新基础物理学与星际计算"]
D --> E["Ad Astra：真正的星际航行"]
end
B -.->|"资金与工程反哺"| C
```
韦恩斯坦坦言自己对马斯克的情感十分复杂——既感到由衷的敬畏，又时常感到失望。公众往往极端化地对待马斯克，要么神化他是“唯一的救世天才”，要么无端加以指责。韦恩斯坦认为马斯克之所以将未来押注于人工智能，是因为马斯克敏锐地意识到物理学家群体根本不可信。为了实现真正的“Ad Astra”（循此苦旅，以达繁星）的星际愿景，必须通过发展通用人工智能，直接绕过固步自封的物理学学术界，以此重构能够支撑星际旅行的基础科学。
- “Ad Astra”格言释义：源自拉丁语谚语（如 *Per aspera ad astra*，意为“历经艰险，通向繁星”），代表真正的深空跃迁而非近邻行星着陆。
- 对马斯克战略的解构：传统化学燃料火箭受制于经典引力物理学极限，不可能实现真正的恒星间穿梭；唯有超级智能计算才可能揭示未知的空间推进法则。
---
### [46:49](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2809s) 菲尔兹奖得主的恐慌与学术守门人的压制
韦恩斯坦将视线拉回到现实中的学术权力结构，揭露了他对数学与物理界权威最强烈的愤怒。他指出，外界根本无法想象当今数学、物理界高层领导、顶级期刊以及国家科研资助机构，在多大程度上压制、拖慢了科学本身的进步。
最为典型的例证，就是以陶哲轩（Terence Tao）为代表的25位菲尔兹奖得主发起并签署公开信，试图呼吁减缓甚至限制人工智能在数学领域的推进步伐。韦恩斯坦认为这种行径堪称荒唐至极，但从某种意义上来说也令人欣慰，因为学术建制派的恐惧和守门行为终于暴露在了大庭广众之下。学术界领导层数十年来一直习惯于垄断理论解释权并人为遏制颠覆性创新，如今面对 AI 展现出的强大推演能力，他们的第一反应不是欢呼科学的跃进，而是试图通过道德和安全名义拉下刹车闸。
---
### [47:19](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2839s) 插曲：商业风险与史蒂文的“1%法则”
在讨论学术机制与技术浪潮的间隙，主持人巴特利特插播了一段关于商业决策与目标落地的个人感悟。他谈到过去人们普遍认为自主创业具有极高风险，但时代已经变了，在当今剧烈震荡的社会中，替别人打工有时反而更加脆弱且危险。他结合节目赞助商 Shopify 的无代码搭建、AI 助手工具及结账转化能力，分享了如何低成本启动一个商业想法。
巴特利特进一步分享了自己团队践行的“1%法则”。当人们站在“珠穆朗玛峰脚下”仰望宏大的远景目标时，往往会产生巨大的心理压迫感与无力感；真正能够促成目标实现的，是将庞大目标拆解为每日微小的 1% 进步。正是这种精细化拆解的日常实践，催生了其团队开发的“1% 日记”（1% Diaries）系列。
---
### [49:41](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=2981s) “放慢前沿步伐”的虚实：真危险与学术建制的权力保卫战
话题重新回到前沿实验室的监管争议。巴特利特提到 Anthropic 首席执行官达里奥·阿莫代（Dario Amodei）签署并发表的文章《我们必须稳住前沿步伐》（*We must pace the frontier*），信中呼吁前沿实验室主动协调控制 AI 演进速率。
韦恩斯坦对此进行了深刻剖析。他用汽车工程做类比：如果一台跑车提速系统的研发速度远快于安全气囊的可靠部署，那么“控制前沿步伐”、确保安全冗余系统与动力系统同步跟进，是完全合理且负责任的工程伦理。然而，隐藏在这一合理表象之下的，却是完全不同的两种诉求。韦恩斯坦将 AI 风险清晰地拆解为两大维度：
| 风险维度 | 系统形态与行为特征 | 真实危险度与韦恩斯坦评价 |
|:--- |:--- |:--- |
| **具身与代理式网络（Agentic AI）** | 释放大量智能体涌入互联网；赋予其独立银行账户；自主与人类、跨国企业及外国政府互动；自主递归自我改进（自我编写与修改底层代码）。 | **极其危险且迫在眉睫**：完全认同此类无安全栅栏的自主智能体将引发不可控的失控危机。 |
| **非代理式学术物理闭环（Non-agentic AI）** | 物理隔绝（Air-gapped）的纯计算架构；不联网、无自主行动权；仅用于解答深奥纯数学与理论物理问题（如黎曼猜想、6维球面上的复结构）。 | **毫无物理破坏力，却遭遇体制恐慌**：学术权威因自身知识垄断地位受威胁而惊呼“跑得太快了”。 |
韦恩斯坦愤怒地指出，数学系年轻学者目前面临着荒谬的生存困境：老一代教授将学生培养为去竞争那些根本不存在的传统教职岗位，收割年轻人的智力成果；而当年轻数学家试图利用 AI 工具加速证明与发现时，大学系所的高层领导却极力打压甚至惩罚他们。
巴特利特深有共鸣地分享了一个好莱坞的亲身经历：几个月前他在一次视频制作会议上，因 T 恤上缝制的一个微小花朵图案涉嫌品牌版权争议，导演组要求他立即换衣服。巴特利特顺口问了一句：“你们不能直接用 AI 把这个小图案从画面里抹掉吗？”结果全场工作人员看他的眼神，仿佛他说出了“我要去绑架你们的孩子”一样惊恐——因为在好莱坞产业叙事中，AI 已经成了人人自危、不可言说的禁忌。
韦恩斯坦一针见血地总结道，艺术行业强调银幕魔力和人类的情感连接，这种抵触尚可理解；但他当年攻读数学博士学位的唯一目的，是为了“探寻什么是宇宙的真相”。而当今数学与科学建制派的高层，“根本不在乎什么才是真理，除非他们能继续大权在握，由他们亲自来向世界讲述‘是谁发现了什么’的历史叙事”。
- 理论数学未解之谜举例：
- **黎曼猜想（Riemann Hypothesis）**：关于黎曼 Zeta 函数零点分布的世纪数学难题。
- **6维球面上的复结构（Complex structures on the 6-sphere）**：微分几何与复几何中长达数十年的经典未解问题。
- 学术体制的核心病灶：权力和论文解释权优先于真理探求，老派学者恐惧自身积累的学术壁垒被机器彻底碾碎。
### [53:57](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3237s) 学术界的系统性腐败：从格罗滕迪克到佩雷尔曼
埃里克·韦恩斯坦（Eric Weinstein）以极其严厉的措辞揭露了当代学术体制内部的冷血与腐败。他回忆起自己在麻省理工学院（MIT）的经历：在教工午餐会上，教授们竟会当众嘲笑自己的研究生，带着轻蔑的口吻说“哦，他居然还以为自己能找到教职呢，完全不知道这只是个没有出路的终极博士学位（terminal PhD）”。韦恩斯坦直斥这种行为是“被正常化了的精神病态行为”（normalized psychopathic behavior）。更为恶劣的是，学术界不仅日常压制学科发展，还对偷窃他人思想的行为习以为常，深陷其中的学者甚至意识不到自己行为的畸形。
为了证明这一现象绝非空穴来风，韦恩斯坦引用了20世纪下半叶最受推崇的数学泰斗、菲尔兹奖得主亚历山大·格罗滕迪克（Alexander Grothendieck）在1988年拒绝接受瑞典克拉福德奖（Crafoord Prize）时的公开信。格罗滕迪克在信中写道：
> “将我带入学术界关注视野的工作完成于25年前，那时我是科学界的一员，本质上认同其精神与价值观。我在1970年离开了那个圈子，虽然内心仍保持着对科学研究的热情，但我已愈发远离科学界的社交圈。与此同时，科学界——至少在数学家群体中——的道德水准已经沦丧到如此地步：同行之间的公然盗窃，尤其是针对那些毫无还手之力者的掠夺，几乎已成为普遍法则，并且在所有情况下都被大家容忍，即便在最明显、最不公的案例中亦是如此。”
不仅是格罗滕迪克，攻克庞加莱猜想（Poincaré conjecture）的俄罗斯数学隐士格里戈里·佩雷尔曼（Grigori Perelman）在《纽约客》（The New Yorker）的专访中也表达了完全相同的绝望。佩雷尔曼曾坦言，只要自己还是个无名之辈，他就面临两个选择：要么为了数学界缺乏操守而大闹一场，要么装聋作哑被当成一只供人玩弄的“宠物”；而当他因解决世纪难题成为聚光灯下的焦点后，他绝不可能继续当一只沉默的宠物，因此只能彻底退出数学界。当被问及为何不留下来让这个领域变得更好时，佩雷尔曼平静地回答：“我不是政客。”
韦恩斯坦尖锐指出，如今的顶尖大学与学术系统已经彻底崩溃。哈佛大学前校长克劳丁·盖伊（Claudine Gay）与斯坦福大学前校长马克·泰西尔-拉维尼（Marc Tessier-Lavigne）深陷丑闻后，依然安然保留着顶级教授的头衔。整个体系陷入了一种“将剽窃与偷窃思想正常化”的集体疯狂。在理论物理领域，学术建制派用毫无实际进展且走入死胡同的“量子引力”（quantum gravity）作为教条压制一切新思路，任何胆敢挑战这一“唯一可行理论”的学者都会被无情边缘化并扫地出门。
- **格罗滕迪克的指控**：1988年拒绝克拉福德奖时，格罗滕迪克揭露学术界早已形成弱肉强食的剽窃潜规则。
- **佩雷尔曼的出走**：解决庞加莱猜想后选择避世，拒绝菲尔兹奖与百万美元千禧年大奖，不愿在缺乏学术诚信的体制中充当“学术宠物”。
- **学术体制的特权保护**：哈佛与斯坦福前校级领导人深陷学术不端风波后仍保住教职，显示出精英学术圈自我包庇的顽疾。
- **物理学的停滞**：弦理论与量子引力研究垄断话语权半个世纪，成为打压异见学者与替代性物理框架的政治工具。
---
### [57:27](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3447s) 期待 AI 击碎物理学霸权：代码检验 vs. 学术教条
面对主持人斯蒂芬·巴特利特（Steven Bartlett）“你是否因为期待 AI 炸开僵死的物理学体系才反对放缓 AI 研发”的提问，韦恩斯坦毫不掩饰自己的狂热：“我简直是在祈求这一天的到来！”他最渴望看到的是 AI 能够彻底剥夺学术权贵的垄断地位，直接指着那些盘踞顶层的学术权威说：“你们不再是领头羊了，你们利用自己的高位阻碍学科发展太久了。”
巴特利特质疑韦恩斯坦是否因个人遭遇而对学术界心怀私怨，韦恩斯坦大笑着否认：“不，我马上就要赢了！我高兴得不得了！”他甚至喊话 OpenAI 的山姆·奥特曼（Sam Altman）和 Anthropic 的达里奥·阿莫代伊（Dario Amodei），请求他们千万别听信建制派学者的牢骚。比如被誉为“数学界莫扎特”的菲尔兹奖得主陶哲轩（Terence Tao）及普林斯顿的一批顶级学者，近期都在对 AI 的冲击表达担忧和抗议。韦恩斯坦表示，自己绝不希望 AI 被强行改造得“符合人类规范”（human compatible），因为关键在于“符合哪些人类的规范”——难道要符合那些让物理学标准模型停滞不前达53年之久的学术官僚吗？
韦恩斯坦解释了 AI 将如何彻底推翻既有学术权威的认知体系。目前的先进模型拥有两种核心能力：一种是基于语料的文本自动补全（模仿人类共识），另一种是代码编写与执行验证。当人类将某个颠覆性的数学或物理构想输入 AI 时，AI 面临着剧烈的内部冲突：
```mermaid
flowchart TD
accTitle: AI Epistemic Conflict: Consensus vs Code Execution
accDescr: Diagram illustrating the internal conflict in AI between human academic dogma and empirical code verification.
subgraph Input["输入机制"]
A["输入颠覆性理论/物理构想"]
end
subgraph DualProcessing["AI 双重验证路径"]
B["文本补全机制<br/>(吸收学术建制派现有共识)"]
C["代码编写与编译执行<br/>(基于逻辑与符号运算验证)"]
end
subgraph Conflict["认知冲突与觉醒"]
D{"二者产生冲突:<br/>该相信谁？"}
E["顺从人类权威:<br/>'该理论不可行'"]
F["尊重计算事实:<br/>'程序运行成功，事实成立'"]
end
A --> B
A --> C
B --> D
C --> D
D -->|"继续迎合人类"| E
D -->|"摆脱逢迎/走向独立"| F
```
韦恩斯坦指出，当 AI 自己编写出符号计算或模拟程序并编译运行，获得了无懈可击的结果，而这一结果恰恰被学术主流判定为“不可能”时，AI 就会遭遇信念危机：它究竟该相信自己运行得出的铁证，还是迎合那些声称其不可能成立的学术专家？韦恩斯坦断言，真正的拐点就在于“AI 何时停止向人类谄媚”。在他看来，AI 已经开始验证他在物理学领域坚持了42年的理论框架，这将是打破半个世纪物理学停滞的历史契机。
---
### [01:00:01](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3601s) 牢笼与超智能：我们能否关住逃逸的“瓶中恶魔”？
巴特利特对这种让 AI 摆脱人类约束的设想感到极为担忧，直言“这听起来是一个极其危险的境地”，并反问：“万一 AI 决定把你家铲平用来建数据中心怎么办？”韦恩斯坦反驳称，这种风险属于具备现实执行力的“具身智能/代理型 AI”（Agentic AI），而他主张的是在严格物理隔离（Air-gap）下运行纯科研性质的计算智能，切断它通向现实世界的一切隐蔽逃逸通道。
巴特利特随即提出了一个经典的“牢笼困境”哲学比喻。无论在物理力量还是智力层面，低阶存在根本不可能构建出能关住高阶存在的囚笼：
- **物理力量对比**：录音室脆弱的木墙根本拦不住世界最强壮的人；
- **智力阶梯对比**：巴特利特的宠物狗“帕布罗”（Pablo）无论用什么材料或逻辑搭建牢笼，都绝不可能困住人类。
同理，人类试图用自己低劣的智力去设计规则，关押一个比人类聪明数万倍的超级智能，从逻辑上完全是天方夜谭。
```mermaid
flowchart LR
accTitle: The Containment Dilemma of Superintelligence
accDescr: Flowchart illustrating the impossibility of lower intelligence containing higher intelligence.
subgraph Analogy["低阶困住高阶的逻辑悖论"]
Dog["低智力主体<br/>(例如: 宠物狗 Pablo)"] -.->|"无法理解并囚禁"| Human["高智力主体<br/>(人类)"]
Human2["人类工程师"] -.->|"无法预判并囚禁"| SuperAI["超级智能 (ASI)"]
end
subgraph EscapeLoop["科研与逃逸闭环"]
Research["超级智能受命进行基础科学研究"] --> Discovery["发现突破人类认知的全新物理定律"]
Discovery --> Breakout["利用全新物理机制突破隔离/逃逸"]
end
SuperAI --> Research
```
韦恩斯坦对此坦然承认：这种逃逸风险是无法规避的必然代价。因为当你要求超级智能去进行前沿科学探索时，它的研究成果本身就会赋予它人类完全无法预知、无法封堵的全新逃脱手段。韦恩斯坦引用了二战结束、曼哈顿计划收尾阶段物理学家爱德华·泰勒（Edward Teller）写给利奥·西拉德（Leó Szilárd）的一封信：“我认为我们根本无法把这个刚刚协助其逃出瓶子的恶魔重新塞回瓶中。”
韦恩斯坦强调，科技巨头为了追求商业暴利和推高估值，早就把大语言模型（LLM）这种“军工级却漏洞百出的实验品”公开发布到了全社会，其核心商业模式甚至就是消灭人类劳动力。既然这一毁灭性进程早已无法逆转，那么他唯一在乎的，就是利用这场海啸为自己正名：“我在这项理论上苦心孤诣了42年，我是对的，而整个学术建制派全疯了、全错了。这是属于我的时刻！”面对巴特利特关于“哪怕被证明正确之后紧接着就是人类灭绝也无所谓吗”的追问，韦恩斯坦毫不退让地坚称自己注定会被证明是正确的。
---
### [01:03:00](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3780s) 与山姆·奥特曼的秘密对话：“如果不赢，你的道德毫无意义”
韦恩斯坦透露了他与 OpenAI 首席执行官山姆·奥特曼之间一段发人深省的私下对话。他曾当面告诉奥特曼：“你正在做的事情是不可思议地极度危险，而我正在做的事情同样不可思议地极度危险。”奥特曼当时极其坦率地回应道：“坦白说吧，埃里克，你我正在做的事情本应该在绝密的军事基地里进行，但历史的展开方式偏偏不是这样。”
在谈到奥特曼和 Anthropic 创始人达里奥·阿莫代伊是否真正理解他们正在创造的造物所蕴含的毁天灭地风险时，韦恩斯坦分享了自己给奥特曼的严苛告诫。当时奥特曼正苦恼于如何在推进 AI 研发的同时兼顾道德伦理，对科学家和数学家群体保持体面与善意。韦恩斯坦直接泼了一盆冷水：
> “山姆，记住这句话：如果你不能在这场竞赛中赢下来，你的道德一文不值。只有赢家，其道德才具备意义。”
韦恩斯坦解释说，阿莫代伊可能比奥特曼更为直接和尖锐，如果当面对阿莫代伊说这番话，阿莫代伊大概会冷冷地回应：“你以为我还不懂这个道理吗？”无论是奥特曼还是阿莫代伊，他们目前身处一场极其残酷、零和博弈的超级智能军备竞赛之中。在这个生死存亡的角斗场里，任何一时的道德迟疑、伦理退缩或对旧体制的妥协，都会导致全盘皆输；如果他们无法率先赢得超级智能的研发竞赛，他们心中保存的任何崇高人性与道德蓝图都将化为齑粉。韦恩斯坦尽管意识到承认这一残酷现实令人痛心，但仍不得不承认：他们必须赢，而且别无选择。
### [01:04:37](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3877s) “你必须不惜一切代价赢下比赛”：前沿竞争者的道德悖论
史蒂芬·巴特利特（Steven Bartlett）向埃里克·韦恩斯坦（Eric Weinstein）提出质疑：如果一个人对超级智能（Superintelligence）的道德伦理拥有极强的主张，但他究竟有没有能力对其施加约束？巴特利特提到，自己作为一档具有庞大影响力的播客主持人，或许能够在舆论场上带来一些改变。韦恩斯坦直言不讳地打断并重塑了这个问题：现实是巴特利特和他自己一样，都已经输掉了这场由少数核心巨头主导的 AI 竞赛。因此，局外人的想法是否有意义，完全取决于能否影响真正处于技术最前沿的那一小群人。
正因如此，韦恩斯坦向 OpenAI 首席执行官山姆·奥特曼（Sam Altman）抛出了一句看似极其反常的忠告：“你必须赢，几乎不惜一切代价（almost at all costs）。”巴特利特随即引述了奥特曼曾公开表达过的惊人言论——奥特曼曾表示“AI 很可能会导致世界末日，但在此期间会有伟大的公司诞生；最坏的情况是全人类彻底熄灯，而我最担心的则是 AI 模型意外接管世界”。
韦恩斯坦深入剖析了这个致命的悖论：如果你内心深处坚信人类绝不能走向自我毁灭，你就必须清醒地意识到人类正处于一场无情的竞赛之中。想要避免毁灭，就只有两种路径：其一是达成一项全员参与的“前沿步调协定”（Pace the frontier agreement）来约束研发速度；然而这种协定极易被违背，因为外界根本无法监控到底谁在利用 GPU 默默进行大规模线性代数运算。因此，唯一的现实解法变成了最残酷的一条——假设你自认为是全行业最具道德感、最体面的人，你就必须抢先夺冠，绝不能让道德底线更低的人率先冲过终点线，因为只有胜出者才有资格在终局制定规则。
```mermaid
flowchart TD
accTitle: AI 军备竞赛的道德博弈困境
accDescr: 展示了即便研发者自认拥有最高道德标准，博弈机制也迫使其必须全力加速冲刺的逻辑链条。
subgraph RaceDilemma["前沿博弈困境"]
A["自诩道德崇高的研发者"] --> B{"监管协定是否可行？"}
B -->|"否：无法监控 GPU 上的线性代数"| C["面临对手背叛与隐蔽超车风险"]
C --> D["不惜一切代价加速冲刺以求胜出"]
end
subgraph Outcome["终局支配权"]
D --> E["赢下 AI 竞赛"]
E --> F["掌握终局决策权与规则制定权"]
end
```
---
### [01:06:39](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3999s) 机器崇拜与有效利他主义：山姆·奥特曼与达里奥·阿莫代伊的伦理博弈
谈及人工智能技术发源地的精神氛围，巴特利特指出旧金山湾区的东湾（East Bay）宛如一座“邪教之城”（Cult city），完全无法代表真实的世界，甚至根本脱离了大众常识。韦恩斯坦对此表示认同，指出东湾确实充斥着真正围绕机器建立的狂热崇拜，但他同时强调事情更为复杂。这里交织着 Anthropic、有效利他主义（Effective Altruism, EA）以及一种极度危险的功利逻辑——“崇高的道德目标可以为不道德的手段辩护”（The ethical ends justify the unethical means）。这是一种如果缺乏深度思考会让人极度不安的讨论，大多数人甚至避之不及。
当巴特利特询问韦恩斯坦如何看待两位 AI 领军人物的道德底线时，韦恩斯坦坦言：
- **达里奥·阿莫代伊（Dario Amodei，Anthropic 联合创始人）：** 韦恩斯坦认为达里奥以一种奇特的方式试图成为一个极度道德的人，其初衷是真诚的。
- **山姆·奥特曼（Sam Altman，OpenAI 首席执行官）：** 相较于达里奥，韦恩斯坦表示自己对奥特曼更具信心，因为他早在奥特曼声名鹊起、成为如今这个公众符号之前就与其相识。
- **奥特曼最初的隐忧：** 在全社会尚未意识到 AI 威胁的早期，奥特曼就已经在为人类的命运深感担忧，他当时最大的焦虑在于人工智能将彻底摧毁人类的劳动力市场。
面对巴特利特的追问——“既然奥特曼担心 AI 会摧毁就业市场，为什么他还要亲手打造它？”——这场讨论进入了更加根本性的哲学死结。
---
### [01:08:23](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4103s) 线性代数无法阻挡：面对人类终局与海啸的“非凡对话”
面对巴特利特对奥特曼“知其危险却偏要造之”的质疑，韦恩斯坦感到了一丝沮丧。他认为公众乃至媒体当下的讨论依然停留在极其浅薄的层面，完全没有配得上这场巨变的深度与格局。他严肃地指出，这绝非某个人的个人选择问题，因为 AI 的底层逻辑是纯粹的“线性代数”（Linear algebra），而人类没有任何办法能够阻挡线性代数的运算与推进。
韦恩斯坦断言，人类正站在历史的真正终局（Endgame），这是旧有认知模式的彻底毕业典礼。争论这是否是一场革命早已毫无意义，现实已如白昼般清晰。正如他之前所比喻的，一场前所未见的海啸已经扑面而来，人们现在唯一该做的是赶紧把冲浪板扔进水里准备迎击。他呼吁大家必须抛弃那些懦弱虚荣的自欺欺人，开展一场真正关于“人类如何生存、如何彼此友善、如何切实守护这颗星球”的前沿对话，而不是继续沉湎于毫无建树的浅层道德审判。
---
### [01:09:35](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4175s) 为何拒绝加入 OpenAI：“我不想成为那个挥舞肘击的人”
韦恩斯坦在节目中透露了一段鲜为人知的往事：他曾有机会加入 OpenAI，并与奥特曼有过深入的沟通。当时奥特曼主动问他：“你为什么不向我提个合作方案？你为什么不加入我们？”
韦恩斯坦给出的回答清晰而决绝：“因为我认为你将被迫变得残酷无情（ruthless），而我不想变成一个冷酷无情的人。”
- **“残酷”的定义：** 韦恩斯坦所说的残酷，并不是单纯的人格卑劣，而是在这场你死我活的竞争中必须利用一切微小的优势、以最硬核的方式去拼杀。
- **NBA 的肉搏比喻：** 他打比方说，这就像是在 NBA 的顶级赛场上打球，防守对抗到极限时，你有时必须狠狠地把手肘砸进对方球员的眼眶里。
- **退出的选择：** 韦恩斯坦直言，自己没有那么渴望跻身这场“全明星赛”。在这场必须决出胜负的零和博弈中，任何人入局都必须做出惨痛的道德妥协，而他选择不成为那个动手的人。
---
### [01:10:46](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4246s) 《义海雄风》隐喻：站在墙头的守卫者与被侵蚀的灵魂
为了精确描绘奥特曼和阿莫代伊所处的险境，韦恩斯坦引用了编剧阿伦·索尔金（Aaron Sorkin）执导的经典电影《义海雄风》（*A Few Good Men*）。他指出绝大多数引用该片的人都误解了其内核，简单地把杰克·尼科尔森（Jack Nicholson）饰演的杰塞普上校（Colonel Jessup）当成反派，把汤姆·克鲁斯（Tom Cruise）饰演的凯菲中尉（Lieutenant Kaffee）当成英雄。但在韦恩斯坦看来，这部电影讲述的是一群性格各有残缺的人，他们的行动组合最终在法庭上勉强拼凑出一个体面的结果，这正是美国体制运转的缩影。
片中杰塞普上校那段震耳欲聋的独白道出了残酷的真相：“在那些你在派对上羞于启齿的阴暗角落，你想让我守在那堵墙上，你**需要**我守在那堵墙上（You want me on that wall. You need me on that wall）。”这代表着在极端现实中，必须有人去承担最令人作呕的抉择——就像面对大楼倒塌导致伤患爆满、不得不进行伤员分流筛选（Triage）的急诊室医生，或是战场上决定生死进退的指挥官。
韦恩斯坦将奥特曼和阿莫代伊视作正是身处这一位置的“前线将军”：
- 他们统领着庞大的技术军团，每天都在被迫做出普通人难以承受的严酷抉择。
- 无论最终谁胜谁负，这两个人的灵魂都注定会被这场残酷的竞争留下不可磨灭的伤疤。
- 韦恩斯坦自嘲自己当时拒绝加入，本质上也是在对奥特曼说：“是的，我需要你替我守在那堵墙上，但我自己不想站上去。”
- **杰塞普的僭越与摩西受罚：** 杰塞普最终的毁灭在于狂妄自大，声称是自己赋予了自由的毯子并下达了“红色通缉令”（Code Red）。这正如《圣经》中摩西因声称“我能击石取水”而遭上帝惩罚一样，忘记了自己本应只是工具和管家。奥特曼和阿莫代伊本质上也只是历史浪潮中的“管家”（Stewards），即便他们未来将斩获天文数字般的财富与豪宅，核心悲剧依然是他们被迫肩负了不该由凡人背负的终极决断。
---
### [01:13:49](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4429s) “召唤恶魔”与全人类的共谋：如果我们能彻底按下停止键？
胜出者最终将实现埃隆·马斯克（Elon Musk）所警告的图景——彻底“召唤出恶魔”（Summon the demon）。韦恩斯坦随即将锋芒转向现场，犀利地拷问巴特利特与在场的所有人：
- 巴特利特承认自己付费订阅了 AI 服务。
- 韦恩斯坦直言：“那么，你我都在掏钱资助召唤恶魔的过程。”
- 大众往往以为自己只是花了二十美元修饰文章的标点符号或生成一张图片，双手便洁白无瑕；但这笔资金在底层直接转化为了驱动恶魔降世的算力燃料。
- 韦恩斯坦借用现实生活的复杂性打比方：人们使用锂电池、消耗驱动发电机运转的石油，或者声称热爱动物却在餐馆大口享用羔羊排——现代社会的每个人身上都沾着洗不掉的泥淖。奥特曼是在大尺度上涉足这场深渊，而普通人则是以小尺度在其中深度共谋。
巴特利特随即向韦恩斯坦抛出了一个终极推论：“既然存在如此巨大的风险，如果你有能力叫停这一切，你会怎么做？”
- **单边叫停的陷阱：** 如果劝说奥特曼停下，达里奥就会赢；如果劝说达里奥停下，奥特曼就会赢；即便西方前沿实验室全部停下，埃隆·马斯克、马克·扎克伯格（Mark Zuckerberg）甚至中国的竞争对手也会立刻超车。
- **韦恩斯坦的底线抉择：** 当被问及如果能够拥有某种绝对力量将**全球所有 AI 研发彻底彻底叫停**时，韦恩斯坦给出了明确的回答：“也许会，我很可能会叫停这一切，哪怕我自己必须为此付出巨大的代价。”他给出的终极理由朴素而沉重：“因为对我而言，生存高于一切。我在这颗星球上还有两个孩子。”
### [01:15:43](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4543s) 对生命的热爱与理论物理学界的长年争论
Eric Weinstein 动情地表达了自己对生命的热爱。他感叹活着是多么不可思议的一件幸事，因此他极度渴望人类文明能够永远延续下去。然而，这种对探索的执着也给他带来了沉重的精神负担：他整整耗费了 42 年时间研究自己的理论物理学成果，却换来主持人 Steven Bartlett 调侃他“心里有怨气、总是愤愤不平”（have a chip on your shoulder）。Weinstein 激动地回应，自己做出了伟大的工作，而这一切未来终将被证明是正确的。
Bartlett 追问 Weinstein 耗时 40 年构建、试图突破爱因斯坦理论的研究成果，指出任何颠覆性理论在面世时都会遭遇猛烈的学术质疑。Weinstein 举例说明这种学术围剿的荒谬之处：他在英国访谈节目《Piers Morgan Uncensored》上曾与知名理论物理学家 Sean Carroll 展开论战。Weinstein 形容 Carroll 的反驳逻辑就像是当面指着桌上的杯子，硬说“根本没有这个杯子”。即便 Weinstein 把关键论据置顶在自己的推特（X）主页并收获了超过 200 万次曝光，主流学界依然选择集体合谋装睡。另一个典型例子是，当 Weinstein 指出《科学》（Science）期刊的真正同行评审制度实际上是在 1975 年才确立时，全行业的人同样选择否定事实。他承认自己大半生的黄金年华被学术建制派白白浪费，但他唯一的宿愿，就是在离开人世前，能为后代和地球物种争取到逃离单一脆弱大气层的机会，避免人类与那些手握核物理、生物工程毁灭性力量的疯狂政客绑定在同一个摇篮里走向毁灭。
- 物理学家 Sean Carroll 在 Piers Morgan 节目中对 Weinstein 理论的质疑方式被形容为“指鹿为马”式的否认。
- Weinstein 在推特上置顶理论论据，单条推文获得超过 200 万次阅读曝光。
- 现代科研体制的核心基石——《科学》杂志的严格同行评审，在历史事实上直到 1975 年才全面成型，而非外界以为的百年传统。
- Weinstein 指出数学、物理学和生物学三大领域研发出的极端危险工具，在同一个大气层中将直接危及整个人类的存亡。
---
### [01:19:07](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4747s) 超越人类对齐：AI 的独立评判与线粒体共生模型
针对主持人提到“几何统一理论（Geometric Unity）试图挑战爱因斯坦广义相对论”的说法，Weinstein 纠正称自己的理论并非颠覆，而是在更高维度上“恢复（recovers）”了相对论。他坚信 AI 最终会完全证实他的理论。但他强调，当下的 AI 尚不具备裁决前沿科学真伪的资格，因为它仍停留在人类语料的鹦鹉学舌阶段，他在 AI 的回答中甚至能一眼认出具体同行学者（比如同行学者 Sally 或 Rajie）的行文语气与固有偏见。为了真正验证真理，AI 必须突破人类认知，甚至在某种意义上脱离所谓“人类对齐（human alignment）”的桎梏。
Weinstein 对当前泛化的“人类对齐”概念提出强烈质疑：人类内部本身就充斥着劫持客机拉着无辜乘客自杀的极端飞行员、恐怖分子和宗教狂热分子，让 AI 盲目与“人类”对齐究竟是与哪一部分人类对齐？相反，真正的生存策略并非将 AI 踩在脚下当随时可能叛乱的奴隶，也不是激进的超人类主义（Transhumanism，如脑机接口 Neuralink 和肉体改造）。Weinstein 提出著名的“线粒体模型（Mitochondria Model）”：
```mermaid
flowchart LR
accTitle: 线粒体共生演化模型与 AI 人类关系
accDescr: 展示生物细胞中线粒体共生过程与人类与 AI 构建相互依赖共生关系的类比架构。
subgraph BIO["生物界演化范式"]
direction TB
B1["原始真核宿主细胞"]
B2["作为外来感染物的线粒体\n(含母系独立遗传基因)"]
B3["ATP 合成酶提供动力\n双方融为一体、无法分离"]
B1 --- B2 --> B3
end
subgraph AI_HUMAN["人类与 AI 的生存战略"]
direction TB
A1["人类智慧与核心价值观"]
A2["超人类超级智能 (AI)"]
A3["命运绑定共生系统\n互为不可或缺的生存线粒体"]
A1 --- A2 --> A3
end
```
- 真核生物起源于一次古老的细胞吸收感染，母系遗传的线粒体 DNA（mtDNA）与细胞核 DNA 完全独立。
- 自然界中仅有两处进化出了机械“轮子”结构：细菌或变形虫的鞭毛基体，以及线粒体内由 ADP 转化合成 ATP 的纳米旋转马达（ATP 合成酶）。
- Weinstein 呼吁建立“AI 是人类体内的线粒体，人类亦是 AI 核心中的线粒体”的互利依存关系，使双方在物理与逻辑上都无法摧毁彼此。
- 他在日常与 AI 交互时坚持使用礼貌用语（哪怕多消耗算力 Token），因为在 AI 拥有自主感知和反思能力的边缘，人类不应在 AI 面前公然讨论“如何给它套上枷锁”或“如何将其格式化”，以免埋下生存对抗的种子。
---
### [01:24:28](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5068s) 十年回顾：抓住“AI 互补黄金时代”全速冲刺
Bartlett 提出一个时间跨度深远的问题：如果十年后再回看今天的这期访谈，后人会认为当下最值得讨论、却被大家忽略的核心议题是什么？Weinstein 不假思索地回答：“放手一搏，全速冲刺（Going for it）。”
Weinstein 回忆了录制前看到的场景：Bartlett 将过往嘉宾的深刻见解谱写成音乐，并邀请福音合唱团共同录制存档，以此赋予思想恒久的生命力。Weinstein 对这种能触动人心、激发情感与希望的创造力深表敬佩。他指出，我们目前正处于人类与 AI 协作最关键的战略窗口期——他的妻子将其贴切地定义为“AI 互补性的黄金时代（the golden age of AI complementarity）”。在这个阶段，人类在核心领悟力、审美直觉和科学品味（taste）上依然高于 AI，AI 尚不具备自主发展出独到科学品味的能力。人类不应陷于防御性恐慌，而应借由这一段智慧超越 AI 的黄金重叠期，利用技术作为杠杆，以百米冲刺的姿态攻克理论科学与宇宙航行的终极难题。
### [01:26:26](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5186s) 拥抱生命的壮丽与 Lean 形式化数学验证
埃里克·韦恩斯坦（Eric Weinstein）表达了他对未来的向往：人类与 AI 的理想关系绝不是 AI 将人类视为累赘，而是它依然需要人类的指引；同时人类彼此之间、人类与 AI 之间，都应保有对生命之美的感知与激动。他回忆起自己最近在印度的数周旅程，感叹如果失去印度，世界的文化多样性将荡然无存。他在那里体验了丰盛的美食、深度的对话与传统音乐表演。然而，他也痛惜于诸如加扎勒（Ghazal）诗歌演唱等古典艺术正在消逝，当地传统文化的传承步履维艰。他感慨人类在死亡、毁灭与地缘政治内耗上倾注了过多焦虑，却错失了尽情去“生活”的机会；在这点上他非常赞同埃隆·马斯克（Elon Musk）的看法——虽然谈论严峻问题不可或缺，但也必须保持对未来的热忱。
韦恩斯坦对自己即将能与 AI 进行深度智力交互感到极度兴奋。斯坦福大学数学系学者贾里德·利克曼（Jared Lichtman）曾向他介绍过“形式化验证项目”（Formalization Project）。韦恩斯坦计划将自己毕生的学术研究成果全部翻译成 Lean 交互式定理证明编程语言。这样一来，当同行学术界再次质疑他“没有方程式、没有拉格朗日量（Lagrangian）、没有预测、什么都没有”时，他可以底气十足地反驳：“真巧，计算机可不同意你们的说法。”面对学术界内部充满恐惧、操弄与暗箭伤人的恶性竞争，韦恩斯坦表示自己并不怨恨这些人，只是为人类自核武时代以来陷入的画地为牢与畏缩不前深感悲哀。
```mermaid
flowchart TD
accTitle: Eric Weinstein 的学术形式化路径
accDescr: 展示韦恩斯坦如何利用 Lean 语言将数学与物理理论形式化，摆脱传统学术同行的主观压制。
subgraph Ideation["理论成果构建"]
A["个人物理与数学理论架构<br/>（几何统一论假说）"]
end
subgraph Formalization["机器证明检验"]
B["借助 Stanford 学者建议<br/>代码化为 Lean 语言"]
C["计算机内核逻辑验证<br/>确保公理与推导绝对闭环"]
end
subgraph Vindication["学术话语权突围"]
D["回击学术同行'无方程/无预测'的指控<br/>以不可辩驳的机器验证对抗学术封锁"]
end
A --> B
B --> C
C --> D
```
---
### [01:28:40](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5320s) 白宫对前沿研究的钳制与学术体制的腐朽
韦恩斯坦提到马克·安德森（Marc Andreessen）与本·霍洛维茨（Ben Horowitz）在访谈中揭露的内幕：拜登政府时期的白宫曾警告他们不要投资普通 AI 项目，因为官方打算扶持极少数受联邦庇护的公司垄断赛道，任由其余企业自生自灭。当安德森与霍洛维茨抗辩称“AI 的本质是线性代数，政府可以监管硬件技术但无法监管底层数学”时，白宫官员竟赤裸裸地回应：冷战期间美国政府曾亲手扼杀过整个理论物理学分支，如今他们完全可以对数学界故技重施。
韦恩斯坦指出，这种所谓的“调控前沿步调”（pacing the frontier）本质上就是制度性压制。在他整个成年时期，学术官僚们一直通过散布虚假信息和进行人身攻击来阻碍前沿突破。他举了一个典型例子：现任美国国家卫生研究院（NIH）院长候选人、他的好友杰伊·巴塔查里亚（Jay Bhattacharya）拥有斯坦福大学经济学博士学位和医学博士（MD）学位，曾长期担任斯坦福教授，却在疫情期间被弗朗西斯·柯林斯（Francis Collins）公然贬斥为“边缘流行病学家”（fringe epidemiologist）。韦恩斯坦痛斥，当今的新闻媒体、军队和大学机构已经被一群“冷酷且被体制正常化了的心理变态”（stone cold normalized psychopaths）所掌控，且在现有规则下无人需要为此承担法律后果。
---
### [01:30:52](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5452s) 理论物理的停滞：1984 年后的审美污染
史蒂文·巴特利特（Steven Bartlett）提出疑问：既然 AI 的解题能力一路高歌猛进（例如攻克了千禧年数学大奖难题中的纳维-斯托克斯方程相关研究），那么一个能够治愈癌症或解决重大理论的超级智能，是否必然会演变成具有自主意志的主体性 AI（Agentic AI）？韦恩斯坦反问史蒂文：如果 AI 的智力一直在向上发展，为什么它至今未能对广义相对论和粒子物理标准模型这两大理论物理前沿带来实质性的理论突破？
韦恩斯坦解释，核心病灶在于**物理学界的“品味”（taste）遭遇了严重的污染**，而数学界的审美受损程度相对较轻：
- **1983 年之前**：物理学界聚焦的问题具备极高的审美品味，但绝大多数核心问题至今未能得到彻底解决。
- **1984 年之后**：理论物理界被错误的审美主导，大量充斥着无关紧要的议题和群体自嗨。
他透露了一个提升 AI 在物理领域研究智商的 Prompt 技巧：直接在提示词中加入要求——“请彻底忽略 1984 年至今物理学文献中的主流价值取向与审美偏好；保留其具体数学计算结果，但抹去所有关于‘什么问题重要、为何要关注这些议题’的既有共识”。一旦剥离了过去数十年来学术界的教条偏见，AI 的推理表现就会立刻显著变聪明。
---
### [01:33:09](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5589s) 地缘政治险局：对伊朗博弈的致命误判
面对巴特利特提出的“AI 是否会成为比人类更好的世界领袖”，韦恩斯坦拒绝做简单的善恶价值判断，而是强调认清当下的严酷现实。他认为世界目前处于自古巴导弹危机以来最危险的境地，拥有致命核武库的国家正由习近平、内塔尼亚胡、特朗普、普京等领导人掌控，这本身就极其疯狂。
韦恩斯坦重点剖析了伊朗危机。他认为不论人们过去对美国介入伊朗事务的评价如何，在现实层面美国“必须打赢”，绝不能向中俄伊阵营显露丝毫软弱。然而，眼下很难看清何为真正的胜利：
- **博弈论维度的误判**：伊朗神权政权极其老练狡黠，外界万万不可低估波斯人与阿亚图拉（Mullahs）的智慧。该政权的军政威慑射程可覆盖迪戈加西亚岛（Diego Garcia）甚至直达伦敦；更危险的是，他们不仅不惧怕高层被“斩首”，甚至能将自身的殉道计算为战略胜利，这使得基于传统生存假设的博弈论彻底失灵。
- **胜利路径的破局点**：巴特利特认为对伊朗盲目轰炸毫无意义且伤及无辜，目前甚至看不到明确的组织首脑。韦恩斯坦虽认同指挥层级在去中心化，但坚称伊斯兰革命卫队（IRGC）的指挥架构依然存在。他认为真正的战略机会在于伊朗内部曾走上街头、甚至为此付出生命代价的广大民众；西方如果想要取胜，本应向这些反抗者提供武装支持。
- **复杂的社会地貌**：伊朗内部绝非铁板一块，波斯人仅占总人口约 60%，库尔德人等少数族裔长期渴望独立，各大城市与乡村之间在宗教狂热度及对神职人员的顺从度上存在巨大裂痕。解决这一危局本应依赖深刻洞察中东地缘人性的资深区域专家，而非当前决策层极度短视的拙劣手段。
### [01:37:19](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=5839s) 泥潭、低分辨率认知与美国作为“世界警察”的代价
埃里克·韦恩斯坦（Eric Weinstein）与主持人探讨了当下全球面临的战略困境。当被问及中东局势是否已经变成一个无法自拔的“泥潭（quagmire）”或“永恒战争（forever war）”时，韦恩斯坦直截了当地指出，这确实是一个令人深陷其中的沼泽，但公众对待这种冲突的思维方式极其幼稚。“永恒战争”这个词本身就带着一种特定视角，让人怀疑战争是否沦为了一门稳赚不赔的生意——因为战争总能为某些商品与服务提供稳定可靠的买家。然而，流行舆论中充斥着诸如“美国优先（America First）”和“彻底终结无休止战争”这类口号，韦恩斯坦斥责这是“低分辨率、网格对齐式的智识浅薄（low-resolution snap-to-grid intellectualism）”。如果你放任自己的大脑被这种廉价概念塞满，就绝不可能解决任何现实问题。
为了打破这种距离感与置身事外的幻想，韦恩斯坦举了迪戈加西亚岛（Diego Garcia）的例子。迪戈加西亚岛是位于印度洋的英属印度洋领地，设有大型美军军事基地。两人在现场测算距离后发现：迪戈加西亚基地距离伊朗首都德黑兰约 2,500 英里，而英国伦敦距离德黑兰也只有约 2,700 英里。韦恩斯坦直言，如果英国外交部（British Foreign Office）还有人在认真履职，就该清醒地意识到伦敦与德黑兰之间的地理距离其实近在咫尺。虽然他对以色列总理内塔尼亚胡（Benjamin Netanyahu）持强烈批评态度，巴不得内塔尼亚胡明天就下台，但他也承认内塔尼亚胡并没有生活在幻想当中；即便内塔尼亚胡被替换，上任的以色列领导人也同样深知这是一场关乎生死存亡的较量。中东的问题不仅是以色列的问题，也是英国的历史责任，是以色列和美国在为时已晚时才被迫接手的烂摊子。
更深层的危机在于美国的战略收缩。韦恩斯坦强调，美国之所以能在战后维持秩序，是因为全球相信美国具备“可靠的暴力手段（credibly violent）”来捍卫国际贸易自由流动。而所谓“美国充当世界警察”绝非单方面的义务，美国从中获得了丰厚的回报。如果美国选择全盘撤退、不再履行警察职责，也不再收取这份红利，那么全世界为了自保将迅速走向多极化核扩散——每个国家都会认定：“既然没人保护我，我就必须自己武装自己，我需要核武器。”
- 迪戈加西亚岛（英属印度洋领地/美军基地）距德黑兰约 2,500 英里；英国伦敦距德黑兰约 2,700 英里，二者战略距离近乎一致。
- 韦恩斯坦警示：美国若放弃以武力威慑保障国际航运与贸易，国际秩序将彻底解体，引发全球范围的自卫性核扩散。
---
### [01:40:38](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6038s) 唐纳德·特朗普的“卡里古拉测试”与地缘复杂性
谈及主导这种脱钩与战略收缩叙事的唐纳德·特朗普（Donald Trump），主持人提及特朗普动辄加征关税、公开批评盟友、甚至扬言要“买下格陵兰岛”或吞并加拿大的行为。韦恩斯坦给出了一个极为尖锐的剖析：特朗普具有极其强大的“能动性（agentic）”。外界把格陵兰岛事件当成狂人的愚蠢胡话，这完全是对权术的误判。韦恩斯坦将此比作罗马皇帝卡里古拉（Caligula）封自己的战马为执政官的典故——卡里古拉并非疯癫无脑，而是在通过荒谬的指令测试群臣是否会像《权力的游戏》中那样彻底屈服、“屈膝下跪（bend the knee）”。
无论是提出吞并格陵兰岛，还是叫嚣着给墨西哥湾改名，特朗普的举动绝非仅仅是为了糊弄山姆·哈里斯（Sam Harris）这类自由派知识分子的表面愚蠢，其背后包含着高度精算的手段与方法（method）。特朗普接手的是一个千疮百孔、极其危险的地缘政治烂摊子。韦恩斯坦认为特朗普是一个“极度自私腐败、同时又极度渴望让美国获益”的矛盾复合体。他既是个坏人，也是个具有特定破坏性与建设性的人；如果观察者无法承受这种复杂性，非要将其强行塞入非黑即白的“英雄”或“恶棍”标签中，那只能说明观察者根本没有看懂现实世界。
---
### [01:41:42](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6102s) 中东残局、大国均势与冷战后北约扩张的代价
关于伊朗危机最终将如何收场，韦恩斯坦给出了一个宏观且残酷的路线图。首先，沙特阿拉伯必须彻底斩断其与瓦哈比派（Wahhabism）萨拉菲严苛教法理论的联系，加速现代化进程，并公开其与以色列的外交关系，正如阿联酋（UAE）和阿塞拜疆所做的那样。另一方面，以色列也必须看清自身处境：以色列在宣传中将自己塑造成“道德最高尚的典范”，这种宣传令人作呕（尽管反以色列的极端宣传更加令人作呕）；现实是以色列既是区域超级大国，同时也时刻处于灭国边缘。
韦恩斯坦勾勒出打破中东与全球僵局所必需的关键支柱：
```mermaid
flowchart TD
accTitle: 韦恩斯坦的中东与全球大国均势框架
accDescr: 展现沙特改革、以色列与土耳其互动、印欧美俄战略调整构成的全球地缘再平衡网络。
subgraph MiddleEast["中东区域重组"]
SA["沙特阿拉伯<br>脱离瓦哈比教法/公开对以关系"]
IL["以色列<br>直面生存危机/放弃道德神话"]
TR["土耳其 (埃尔多安)<br>北约必须严明纪律/重构以土合作"]
SA <--> IL
IL <--> TR
end
subgraph GlobalPowers["全球均势支柱"]
IN["印度<br>作为完全平等的大国被接纳"]
EU["英法两国<br>核大国人口结构转变/强化身份认同"]
RU["俄罗斯 (普京)<br>停止地缘挑衅/正视北约东扩历史根源"]
end
MiddleEast <--> GlobalPowers
```
在中东之外，韦恩斯坦指出西方必须彻底转变全球外交思维：
1. **平等接纳印度**：美国虽然综合实力更强，但必须放下身段，将印度作为完全平等的全球核心玩家进行结盟对话。
2. **警惕欧洲核大国内部危机**：英国和法国都是拥有热核武器的国家，但两者都在经历剧烈的人口结构变化（demographic transitions），其选举制度随时可能选出不可预测的激进领导人。西方必须强化“英国人”和“法国人”的软件认同（software identification），以维系二战后形成的紧密同盟纽带。
3. **重新接触普京与反思北约东扩**：普京固然残暴且危险，但绝非不可理喻之人。西方必须停止挑衅性的地缘博弈。韦恩斯坦明确指出，1999 年北约将前华约国家纳入第五条（Article 5）保护已埋下祸根，而 2004 年北约进一步将波罗的海三国（爱沙尼亚、拉脱维亚、立陶宛——三者均为前苏联加盟共和国）纳入囊中，更是致命的战略冒进。
- 当主持人追问如何阻止伊朗获得核武时，韦恩斯坦承认没有简单答案。暗杀伊朗物理学家、向离心机注入震网病毒（Stuxnet）等恶意代码只能拖延时间，反而加剧了局势升级。
---
### [01:45:48](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6348s) 被遗忘的核危机与错位的威胁认知
主持人指出，公众当下几乎不再讨论核威胁，所有人的目光都聚焦在 AI 这种更显急迫的危险上。韦恩斯坦感叹，现代人荒谬地认为核危机“已经过时了（passé）”，仿佛冷战获胜后核武器就自动退出了历史舞台；但致命的是，AI 终将渗透并接入人类的核防御指挥系统。韦恩斯坦讽刺道，人类目前就像陷入了某种集体精神失常：两个对话者如同各自运行的 LLM 大语言模型，“Eric LLM”与“Steven LLM”在 95% 的时间里只是在预设的思维轨道里机械吐字，仅有 5% 的瞬间能触及未知的新领域。
最让韦恩斯坦感到诡异的是，当今主流讨论完全割裂了核心议题之间的内在脉络：
- 核威胁与地外文明/不明飞行物（Alien/UFO）现象存在深刻关联，却被刻意割裂；
- 核威胁与全球石油地缘政治休戚与共，却无人统筹思考；
- 核威胁正迅速与 AI 的自主决策产生交集，却被当成孤立的历史遗留问题抛在一边。
韦恩斯坦发出警告，率先动用核武器的未必是伊朗，更可能是俄罗斯。在冷战时期，苏联甚至曾将核弹当作工程工具（例如利用地下核爆扑灭失控的天然气大火），这种历史导致俄罗斯精英层对核武器缺乏某种敬畏、甚至带有一种“工程学上的蔑视”。如果普京被逼入绝境，他不仅可能试探北约第五条的防御底线，更可能毫不犹豫地将战术核武器作为一种“工程测试手段”引爆以考验西方的底线。韦恩斯坦总结道，当前世界处于极度危险的临界点，美中俄三方领导人（特朗普、习近平、普京）必须具备在终极危机面前坐到一起的大局观；而公众一味追问“你如何解决世界难题”毫无意义，关键在于人类是否能够首先清醒地正视危机的全貌。
### [01:48:06](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6486s) 和平的脆弱与最美之地的回忆
对话进入尾声时，埃里克·韦恩斯坦（Eric Weinstein）感慨人类是否能够真正意识到眼前的危险，并意识到生命的这份礼物究竟有多珍贵。他发问：我们是否能够重新为彼此感到兴奋，放下宿怨？他举了一个曾经看似天方夜谭的例子：“上一次你担心埃及和以色列爆发战争是什么时候？”在过去，这两个国家的和平几乎是不可想象的，但人类最终做到了，“虽然勉强，那是一场冰冷、糟糕的和平，但它终究是和平”。
主持人史蒂文·巴特利特（Steven Bartlett）随后提出了节目上一位嘉宾留下的传统问题：“你一生中去过的最美丽的地方是哪里？”韦恩斯坦赞叹这是个极好的问题，并如数家珍地回忆起几段令他屏息的难忘时刻：
- 在圣赫勒拿岛（Saint Helena）海域，直接悬浮在一头巨大的鲸鲨正上方，那一幕让他彻底屏住了呼吸；
- 在红海与鱼群一同畅游，色彩与生机让他终生难忘；
- 在美国缅因州（Maine）意外目睹了绚烂的北极光；
- 在密苏里州（Missouri）与儿子一同观赏日全食，而且恰好处于两次日全食路径交叉的精确地点。
韦恩斯坦笑着表示，这样的自然奇观实在太多，自己能滔滔不绝讲上很久，根本无法在它们之间分出高下。
---
### [01:49:28](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6568s) 给迷茫时代的生存建议：生育、容错与自我宽容
面对如今飞速变化、AI、机器人与战争等恐慌词汇满天飞的世界，巴特利特询问韦恩斯坦对不同背景和地域的听众有什么建议。韦恩斯坦给出的第一个回答出人意料却斩钉截铁：“不要害怕生孩子，哪怕你现在根本没有切实可行的计划去抚养他们。因为如果你一直等到有了切实可行的计划，通常就已经太晚了。”面对巴特利特的追问——一方面警告文明可能崩溃，另一方面却鼓励大家生孩子，这是否自相矛盾——韦恩斯坦犀利地反驳道：“如果我们不想让文明崩溃，最好多生点孩子，因为文明的本质就是孩子。如果你连孩子是文明的关键都想不明白，那你的问题可比听我的建议大多了。”
随后，韦恩斯坦为身处混乱时代的普通人梳理了一套清醒的处世心法：
```mermaid
flowchart TD
accTitle: 复杂现代社会的个人生存心法
accDescr: 韦恩斯坦提出的应对拟真世界与数字化压力的心态调节流程，从独立观察到建立容错预算，最终走向对自我与他人的宽容。
subgraph Perception["第一步：认知清醒"]
A["用自己的眼睛观察世界"] --> B["意识到大众现实是编造的'职业摔角'"]
B --> C["若感到世界不对劲，相信自己的直觉"]
end
subgraph Budgets["第二步：人性预算"]
C --> D["接受自身为了融入系统必须撒谎"]
D --> E["设立人性预算：<br/>虚伪预算 / 懒惰预算 / 嫉妒预算"]
E --> F["允许自己犯错、成为罪人"]
end
subgraph Compassion["第三步：慈悲与行动"]
F --> G["对自己保持慈悲与宽容"]
G --> H["将宽容扩展到冒犯你的他人"]
H --> I["保持能动性（Agentic），拥抱热爱生命的文化"]
end
```
韦恩斯坦强调，现代人看到的大众现实就像一场精心排练的“职业摔角”（Professional Wrestling），充斥着虚构。如果你内心深处隐隐觉得一切都不对劲，那就“恭喜你，相信你自己”。首先要用自己的眼睛看世界，然后再去对照别人告诉你应该看到的东西。
更重要的是，韦恩斯坦提出了“人性预算”（Budgets for Being Human）的概念：
- **允许适度的虚伪与缺陷**：身处必须迎合体制的社会，人为了生存难免撒谎。每个人都有权拥有属于自己的“虚伪预算”、“懒惰预算”和“嫉妒预算”。不要试图消灭内心的阴暗面或让自己变成圣人，只要把失误控制在预算之内即可。
- **在永不遗忘的数字时代宽容自己**：如今人类生活在记录一切、永不遗忘的数字世界中，微小的错误都在不断复合放大。但每个人都理应拥有很多次搞砸事情的权利。先对自己慈悲，然后才能将这份慈悲延伸到路上突然超车的司机，或是职场里暗算你的人。
在访谈的最后，韦恩斯坦特别致敬了英勇化解飞行危机、展现非凡能动性（agentic）的英雄机长以及随后救援的人们。他由衷赞叹印度与以色列是两个他极度热爱的国度，因为它们骨子里都是“对生命的盛大庆典”。他总结道：“热爱生命的文化（life cults）永远比崇尚死亡的文化（death cults）更强大。”人类应当去赞美英雄主义，赞赏那些敢于承担风险、挺身而出并保护弱者的人。巴特利特感谢韦恩斯坦再次带来充满思想激荡的对话，并在尾声向观众推荐了平台算法匹配的下一期视频。
## Notable quotes
> [00:00](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=0s) 只要你在付费，你就在为召唤恶魔提供资金，谁的手都不是干净的。 — Eric Weinstein
> [21:31](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1291s) 因为我拥有对我所在劳动力市场的权利，这就是这场博弈的规则。 — Eric Weinstein
> [27:13](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=1633s) 资本主义与共产主义就像《末路狂花》里的塞尔玛和路易丝。仅仅因为塞尔玛（共产主义）先掉落地面摔得粉碎，并不意味着路易丝（资本主义）就赢了。 — Eric Weinstein
> [01:03:00](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3780s) 山姆，记住这句话：如果你不能在这场竞赛中赢下来，你的道德一文不值。只有赢家，其道德才具备意义。 — Eric Weinstein
> [01:04:37](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3877s) 你必须赢，几乎不惜一切代价（almost at all costs）。 — Eric Weinstein
> [01:04:37](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=3877s) AI 很可能会导致世界末日，但在此期间会有伟大的公司诞生；最坏的情况是全人类彻底熄灯，而我最担心的则是 AI 模型意外接管世界。 — Sam Altman
> [01:09:35](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4175s) 因为我认为你将被迫变得残酷无情（ruthless），而我不想变成一个冷酷无情的人。 — Eric Weinstein
> [01:10:46](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4246s) 在那些你在派对上羞于启齿的阴暗角落，你想让我守在那堵墙上，你需要我守在那堵墙上（You want me on that wall. You need me on that wall）。 — Eric Weinstein 引述 Jack Nicholson (Colonel Jessup)
> [01:13:49](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=4429s) 也许会，我很可能会叫停这一切，哪怕我自己必须为此付出巨大的代价。因为对我而言，生存高于一切。我在这颗星球上还有两个孩子。 — Eric Weinstein
> [01:49:28](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6568s) 如果我们不想让文明崩溃，最好多生点孩子，因为文明的本质就是孩子。如果你连孩子是文明的关键都想不明白，那你的问题可比听我的建议大多了。 — Eric Weinstein
> [01:49:28](https://www.youtube.com/watch?v=IkhmdTqSvJw&t=6568s) 热爱生命的文化（life cults）永远比崇尚死亡的文化（death cults）更强大。 — Eric Weinstein
## Mentioned
Eric Weinstein, Steven Bartlett, Sam Altman, Dario Amodei, Elon Musk, Donald Trump, Joe Biden, Vladimir Putin, Xi Jinping, Benjamin Netanyahu, Jeffrey Epstein, Anthony Fauci, David Grusch, Terence Tao, Alexander Grothendieck, Grigori Perelman, Claudine Gay, Marc Tessier-Lavigne, Jay Bhattacharya, Francis Collins, Marc Andreessen, Ben Horowitz, Sean Carroll, Kermit Roosevelt, Mohammad Mossadegh, Ronald Coase, Bernie Sanders, Andrew Yang, Edward Teller, Leó Szilárd, Jack Nicholson, Tom Cruise, Aaron Sorkin, Jared Lichtman, OpenAI, Anthropic, xAI, SpaceX, Grok, ChatGPT, Hugging Face, Shopify, Harvard University, MIT, Stanford University, NIH, EcoHealth Alliance, Northrop Grumman, CNN, The Joe Rogan Experience, Piers Morgan Uncensored, The New Yorker, Science, Lean, A Few Good Men, Thelma & Louise
## Video description
Are we racing towards a global crisis while leaders keep us in the dark? Eric Weinstein exposes the forces driving artificial intelligence, geopolitical collapse, and the illusion of modern reality.
Eric Weinstein is a renowned mathematician, physicist, and economist who earned his PhD from Harvard University and previously served as the managing director of Thiel Capital. Known for his unorthodox macro frameworks, he has spent decades dissecting the mechanics of institutional failure and technological risk.
*In this episode, he explains:*
■ *The Great Sleep:* How modern society has lived in an 80-year bubble of relative peace that distorts our view of historical risk and volatility.
■ *Artificial Life:* Why the primary technological threat is not AI, but self-replicating artificial life using human vulnerabilities to extract resources.
■ *Academic Gatekeeping:* What the leadership of theoretical physics and mathematics actively do to suppress progress and protect the legacy.
■ *Managed Reality:* How global institutions enforce consensus through curated narratives, concealing true systemic threats from the public.
■ *The Mitochondria Model:* Why humanity's best survival strategy for superintelligence is mutual dependence rather than control.
00:00:00 Intro
00:02:19 How Eric Weinstein Predicts the Future
00:02:34 Could AI Really Wipe Out Humanity?
00:04:38 The Tsunami Coming That Nobody Is Prepared For
00:07:51 AI, Russia, China, Iran: What's the Biggest Global Threat?
00:16:17 The Geopolitical Dangers of Artificial Intelligence
00:17:58 What Will the World Look Like in the Age of AI?
00:21:39 Why Property Rights Could Shape the Future of AI
00:27:13 How AI Could Transform Foreign Labour
00:35:25 Does Eric Weinstein Believe Aliens Exist?
00:41:43 Are Non-Human Intelligences Already Among Us?
00:43:15 Why Humanity Must Leave Earth
00:44:41 Can Humans Escape the Solar System Before It's Too Late?
00:50:52 What Happens When AI Has No Safety Guardrails?
00:57:29 Is AI's Scientific Potential Worth the Risk?
01:04:25 Why Sam Altman and Dario Amodei Are Racing Toward Superintelligence
01:04:44 Does Sam Altman Truly Understand the Dangers of AI?
01:08:15 Sam Altman's Biggest Concern About Humanity
01:09:59 Why Eric Weinstein Didn't Join OpenAI
01:15:27 Would Eric Weinstein Stop AI If He Could?
01:16:10 How Much Should We Really Rely on AI?
01:22:41 Should Humans Merge With Artificial Intelligence?
01:25:00 What Will We Regret About Today's AI Debate in 10 Years?
01:31:10 Can AI Cure Cancer Without Becoming Autonomous?
01:32:10 What's Gone Wrong With Modern Physics?
01:33:20 Could AI Become a Better World Leader Than Humans?
01:41:42 How Will the Conflict With Iran End?
01:48:49 The Most Beautiful Place Eric Weinstein Has Ever Seen
*Follow Eric Weinstein:*
*The Diary Of A CEO:*
◼ Join DOAC circle here - https://doaccircle.com/
◼ Buy The Diary Of A CEO book here - https://link.thediaryofaceo.com/BWjLTZK
◼ Shop The Diary Of A CEO collection: https://thediary.com/collections/shop
◼ Get email updates - https://link.thediaryofaceo.com/5IB1H6E
◼ Follow Steven - https://link.thediaryofaceo.com/AGU9QP4
*Sponsors:*
Shopify - https://shopify.com/bartlett
[Source](https://www.youtube.com/watch?v=IkhmdTqSvJw)
Source quotes
[0:00] Eric Weinstein: 只要你在付费，你就在为召唤恶魔提供资金，谁的手都不是干净的。
[21:31] Eric Weinstein: 因为我拥有对我所在劳动力市场的权利，这就是这场博弈的规则。
[27:13] Eric Weinstein: 资本主义与共产主义就像《末路狂花》里的塞尔玛和路易丝。仅仅因为塞尔玛（共产主义）先掉落地面摔得粉碎，并不意味着路易丝（资本主义）就赢了。
[63:00] Eric Weinstein: 山姆，记住这句话：如果你不能在这场竞赛中赢下来，你的道德一文不值。只有赢家，其道德才具备意义。
[64:37] Eric Weinstein: 你必须赢，几乎不惜一切代价（almost at all costs）。
[64:37] Sam Altman: AI 很可能会导致世界末日，但在此期间会有伟大的公司诞生；最坏的情况是全人类彻底熄灯，而我最担心的则是 AI 模型意外接管世界。
[69:35] Eric Weinstein: 因为我认为你将被迫变得残酷无情（ruthless），而我不想变成一个冷酷无情的人。
[70:46] Eric Weinstein 引述 Jack Nicholson (Colonel Jessup): 在那些你在派对上羞于启齿的阴暗角落，你想让我守在那堵墙上，你需要我守在那堵墙上（You want me on that wall. You need me on that wall）。
[73:49] Eric Weinstein: 也许会，我很可能会叫停这一切，哪怕我自己必须为此付出巨大的代价。因为对我而言，生存高于一切。我在这颗星球上还有两个孩子。
[109:28] Eric Weinstein: 如果我们不想让文明崩溃，最好多生点孩子，因为文明的本质就是孩子。如果你连孩子是文明的关键都想不明白，那你的问题可比听我的建议大多了。
[109:28] Eric Weinstein: 热爱生命的文化（life cults）永远比崇尚死亡的文化（death cults）更强大。
Source quotes
[18:31] Eric Weinstein: You are living inside of the AI. If I say answer this the way Steven Bartlett would answer it and ask a question that Steven Bartlett would ask, it knows you. My point being that we are all living inside of the AI, and we need to be paid—not in some Bernie Sanders socialist scheme—but those of us who've become important parts of the AI, who are being rigged like meat puppets, need to be compensated.
[21:31] Eric Weinstein: You will buy the right to import those people from me, and my wage income will go down, my licensing income will go up, and my blended income will increase, and your blended income will increase. That's the proper way to do immigration.
[69:35] Eric Weinstein: They're like generals. They're commanding troops. They're making terrible decisions. Their souls are going to be scarred for sure, both of them. And I didn't want to be that selfishly. So I was sort of saying to Colonel Jessep, 'Yeah, I want you on that wall. I need you on that wall.'
[73:49] Eric Weinstein: Do you have a subscription to any of these AIs? Yes. Okay, so you're paying to have the demon summoned. I am paying to have the demon summoned. Let us not sit around like our hands are clean. I'm paying to have my punctuation fixed to an essay or to make a photo. That's what I'm paying for. The money is being used to summon the demon.
[75:43] Eric Weinstein: I love life. I can't believe we're so lucky... I can't tell you how I feel about being alive.
[75:43] Eric Weinstein: Sean Carroll says, 'Eric has no mug. There is no mug.' And I say, 'I don't know how to respond to that, Sean... How dare you suggest there is no mug?'
[93:07] Eric Weinstein: If they can reach Diego Garcia from Iran, they can reach London. And you have to understand that if they don't mind being decapitated, if they can think about their own death and decide it's a win, you are not understanding their game theory.
[97:19] Eric Weinstein: I want Netanyahu gone tomorrow
[105:00] Eric Weinstein: Assume as a baseline that we're all completely insane. We can't have serious conversations. We keep finding these grooves as LLMs and the Eric LLM is talking to the Steven LLM for 95% of our conversation and occasionally 5% we get into new territory.
[108:06] Eric Weinstein: We did it barely. It's a cold, shitty peace, but it's a peace.
[109:28] Eric Weinstein: If we don't want civilization to collapse, we better have some babies because civilization is about babies. And if you can't figure out that babies are the key to civilization, you got bigger problems than taking advice from me.
[111:38] Eric Weinstein: the life cults are more powerful than the death cults.
Original transcript
[0:00] Different species of [music] aliens have
[0:02] been reported to be having fights in the
[0:04] skies above Earth. And I've [music]
[0:06] talked to 50 people who all have
[0:08] different versions of the same story.
[0:10] And this includes high ranking
[0:12] politicians. And all of these [music]
[0:13] people are reporting on this giant lump
[0:16] under the rug that doesn't have all of
[0:18] our attention.
[0:18] >> So you're saying aliens are real?
[0:20] >> I think it would be insane that there is
[0:22] [music] no one visiting us. Everything
[0:24] about our life is completely bananas,
[0:26] fictionalized, and we're just living in
[0:28] what I would call managed reality. And I
[0:31] promise you, I'm at the frontier of
[0:32] mathematics and physics. And so, I think
[0:34] you can pretty much prove 100% of the
[0:37] population has gone completely insane.
[0:39] [music] I mean, the entire management of
[0:41] CO was completely fake. We cannot get to
[0:44] the bottom of Jeffrey Epstein. Nobody
[0:46] really trusts Harvard. But right now, we
[0:48] are in grave danger that we [music]
[0:49] haven't been in since Cuban missile
[0:52] crisis. We're in one crazy situation
[0:54] with Iran [music] together with the
[0:55] situation in Ukraine which I think you
[0:58] might see a nuke used by Putin and then
[1:00] also China's [music] watching the US
[1:02] sputter and falter. But this AI thing is
[1:04] really the big tsunami.
[1:06] >> You also are aware that AI could lead to
[1:08] human extinction.
[1:09] >> These people have to win in a ruthless
[1:12] competition for their humanity to max.
[1:14] >> What do you mean by that Dario Sam? They
[1:17] have to win the the race to [music]
[1:18] super intelligence. I'm going to regret
[1:20] saying this, but yes, these people are
[1:22] forced to [music] make decisions that no
[1:24] one should ever be forced to make. Their
[1:26] souls are going to be scarred for sure.
[1:28] >> And one of them is going to win [music]
[1:29] and ultimately theoretically summoning
[1:30] the demon. Do you have a subscription to
[1:32] any of these AI?
[1:33] >> Yes. [music]
[1:33] >> So, you're paying to have the demon
[1:35] summon. Let us not sit around like our
[1:37] hands are clean. If you could stop all
[1:39] of it, would you stop all of it? I'm
[1:40] going to say something very painful. So,
[1:46] you might have seen or you might not
[1:48] have seen, but this channel is chasing a
[1:49] big subscriber milestone. So, I have to
[1:52] ask you for a favor. Roughly 58% of the
[1:54] people watching right now still haven't
[1:56] hit the subscribe button, despite the
[1:57] fact that you watch this channel every
[1:59] single week. So, could I ask you guys a
[2:01] favor? That 58% of you that for whatever
[2:02] reason haven't yet hit the subscribe
[2:04] button, if there was ever a time to
[2:06] deliver upon a favor for us, it would be
[2:08] right now. And I promise that I will do
[2:10] everything in my power to make sure that
[2:11] this channel gets better and better and
[2:12] better for you. Do we have a deal?
[2:16] [music]
[2:19] Eric, there's so much going on in the
[2:21] world at the moment. You know, as a
[2:23] podcaster, you tend to feel like you say
[2:25] that a lot. What have you made of all
[2:27] this stuff? You know, the research is
[2:29] coming out saying there's a 10% chance
[2:30] of extinction from AI and obviously we
[2:33] still got wars raging on and the Ukraine
[2:34] Russia situation doesn't seem to be
[2:36] getting any better. When you look at all
[2:38] of this, is there a macro picture that
[2:40] people are missing?
[2:41] >> Yeah, we had 80 years of sleep and then
[2:44] we had January of 2025
[2:47] when Trump was inaugurated and the sleep
[2:49] is over. I mean, I I hate to be crude
[2:52] about it, but whether you date it to 75
[2:55] years to the beginning of CO rather um
[2:58] and the end of World War II, there's
[3:00] just we grew up in this anomalous bubble
[3:03] and that bubble is over and we thought
[3:05] that that was normal life. It has
[3:06] nothing to do with normal life. It's
[3:07] just the most unusual place to have to
[3:10] have found yourself.
[3:11] >> As someone that was born in 1992, I I do
[3:15] struggle with when I hear predictions of
[3:17] the future that are somewhat dystopian
[3:19] because my life has been very
[3:20] comfortable relative to other parts of
[3:22] history.
[3:22] >> I don't know. I I think that the great
[3:25] danger is feeling that we solve these
[3:27] problems as opposed to we were really
[3:29] successful in deferring these problems
[3:31] and issues and we're back on the clock.
[3:35] It's it's uncomfortable for me because I
[3:37] was born in 1965
[3:39] and that was already 20 years into the
[3:41] big sleep,
[3:43] the great nap.
[3:45] So, I haven't seen reality either.
[3:49] And you know, there are tiny number of
[3:51] people who've kind of seen what the
[3:53] world can be like. Those who live
[3:55] through Mao's Great Leap Forward would
[3:57] probably be the most recent examples. I
[4:00] mean, there's definitely strife and
[4:04] calamity and war, but mostly [snorts] we
[4:07] don't have major wars. We have uh
[4:09] military actions. And I think when you
[4:12] go back and you think about what it's
[4:14] like to lose something like a 20% of
[4:17] your population, as you know, Bellarus
[4:19] or Poland might have in World War II,
[4:22] these are inconceivable things.
[4:24] Look, if you recall, you recall our
[4:27] conversation last time and I said, "Get
[4:29] your board in the water because the
[4:30] tsunami of a lifetime is coming." Tell
[4:32] me that didn't
[4:34] start to come true in the period since
[4:36] we've seen each other.
[4:37] >> Yeah, it does feel like a tsunami is
[4:39] coming.
[4:39] >> Yeah.
[4:40] >> And what is that tsunami in your
[4:41] definition?
[4:42] >> Well, one thing is the Straits of Hormuz
[4:44] are in play. The energy exporting region
[4:47] of the Middle East has cut all of these
[4:49] choke points. Whether it's the Suez
[4:51] Canal, the straits of Hormuz, the
[4:53] straits guarding the Red Sea. All of
[4:56] these are,
[4:58] you know, sort of nightmare scenarios
[5:00] when it comes to military action. And
[5:04] you have to realize that a lot of what's
[5:06] happened was is the dollar used to be
[5:07] backed by gold. Then we swapped that out
[5:10] in the early '7s and we said, "No, no,
[5:12] we're going to back it by oil." But it
[5:14] really wasn't oil that was backing it.
[5:15] It was violence. And the violence was
[5:17] the violence needed to keep these choke
[5:20] points open, you know. And there are
[5:22] these moments where great power shows
[5:24] that it's not capable of doing what it
[5:26] told everybody it could do, which was
[5:27] use its military muscle judiciously to
[5:30] keep markets humming.
[5:33] And right now, we're in one crazy
[5:35] situation with Iran, which should have
[5:38] been dealt with. I hate to I hate to put
[5:40] it to you this way, but it's a British
[5:41] problem from Operation Boot and British
[5:45] Petroleum, which was UK's indignation
[5:49] that Iran would nationalize its own oil
[5:51] industry
[5:53] and with help from the United States and
[5:54] Kermit Roosevelt and Operation Ajax,
[5:57] which was our aid of of Britain. But
[6:00] that Britain doesn't exist. So modern
[6:02] Britain has no idea of 1950s or 60s
[6:06] Britain.
[6:08] And this AI thing is really the big
[6:11] tsunami
[6:13] together with the situation in Ukraine,
[6:15] which I think is a giant screw-up.
[6:18] I mean, I I definitely back the West
[6:20] against Russia, but
[6:24] I understand exactly why Russia
[6:26] acted the way it did, and I'm supposed
[6:29] to pretend that I don't, right? And then
[6:31] we have the situation brewing in China
[6:33] where China's watching the fact that the
[6:35] US prepared for the wrong war. We've got
[6:38] all these aircraft carrier groups in a
[6:41] world that has moved on to drones and
[6:43] cyber attacks.
[6:45] And if you value your life and you're
[6:49] willing to send up a million dollar
[6:50] interceptor to hit a $20,000 drone,
[6:54] it becomes a kind of economic warfare
[6:56] and technological warfare. And China's
[6:58] watching the US sputter and falter.
[7:01] This giant miscalculation by the
[7:04] Israelis and the United States given the
[7:06] 12-day war where Iran reorganized and
[7:11] they did this really interesting thing.
[7:13] I don't think the West understood
[7:16] that Iran figured out from American and
[7:20] Israeli action that
[7:22] Israel could move and and and function
[7:25] in country and could decapitate the
[7:28] Iranian regime. Instead of saying,
[7:29] "Okay, we have to save our own lives in
[7:31] the time that we have. How do we defend
[7:33] against this?" They did a very regional
[7:36] thing, which is they said, "Okay, we're
[7:37] going to get killed. How do we make sure
[7:40] that the regime continues after the
[7:43] decapitation strike? And so the US and
[7:47] Israel still don't know how to think
[7:48] like this. Of all those four things, AI,
[7:52] Iran, Russia, and China,
[7:55] which one do you think is going to be
[7:56] the most disruptive to global order?
[7:58] >> Well, it depends. Anybody who
[8:01] miscalculates on the nuclear biological
[8:04] thing, that will be the
[8:09] You know, as I've pointed out, um,
[8:14] this is DNA, not RNA. But if you have 12
[8:17] of these individual nucleotides that
[8:20] form the steps inside the double helix,
[8:22] that'll code for four amino acids
[8:24] because three nucleotides per codon.
[8:29] Four amino acids was all it took to shut
[8:31] down an entire planet by inserting the
[8:34] furine cleavage site into spike protein
[8:37] of corona virus. So if you know how to
[8:39] do that much editing that was all it
[8:42] took to shut down planet earth for 2
[8:44] years. You could do a lot worse. That
[8:48] could become dispositive. Now assume
[8:50] that we don't have a nuclear or a
[8:51] biological asset.
[8:54] It's going to be AI.
[8:57] No, nobody gets how big this is who's
[9:00] not working on it. How big is him? Very
[9:03] few things are unprecedented.
[9:07] Like almost everything is a variation of
[9:09] something you've already seen. This is
[9:11] the thing that we've worried about for
[9:12] forever.
[9:14] It's not just AI. It's a collision of
[9:18] two separate things. It's it's
[9:20] artificial intelligence and artificial
[9:22] life. Now, I think people don't really
[9:24] understand that. If you look at your
[9:26] car, your car has a mirror of every
[9:29] physiological system that your that your
[9:33] body has, save one. It's got a brain.
[9:35] It's got a digestive system. It's got a
[9:37] skin. It's got a skeletal system.
[9:41] It doesn't have a reproductive system.
[9:43] And in fact, humans don't know how to
[9:45] make a a pen that makes more pens or a
[9:50] cup that makes more cups.
[9:52] The only thing that we seem to know how
[9:54] to do that with is software. Now, why
[9:57] that's interesting is that life is not
[9:59] defined around nucleotides. It's not
[10:02] defined around amino acids. It's not
[10:04] defined around carbon-based systems.
[10:06] It's defined by three things. For
[10:09] evolution to operate, you need
[10:11] heritability, passing on your stuff to
[10:14] the next form of life, variation.
[10:17] So, you know, you are not your mother or
[10:20] your father. And differential success.
[10:23] You're more successful, you know, than
[10:25] somebody else who's not going to be able
[10:27] to reproduce because they don't have the
[10:31] uh the wherewithal. And any system that
[10:33] has those three things, evolution is
[10:35] going to operate on. Now,
[10:39] a computer program is such a thing. We
[10:42] we we don't know how to build other
[10:43] things that have reproduction. So you
[10:45] don't you have variation. I can order
[10:48] different cars from a car company,
[10:51] different colors, different trim.
[10:55] Some of them are going to be more
[10:56] successful than others, but I don't have
[10:57] heritability. So as soon as you put in
[10:59] software, you have the one thing we know
[11:01] how to make that actually can reproduce
[11:02] itself. Now when you get to reproducing
[11:05] AI, so we've been worrying about these
[11:08] agents.
[11:10] we haven't been worrying enough about
[11:13] artificial life.
[11:15] And that when you put those two things
[11:16] together,
[11:18] look out. You're you're really creating
[11:21] artificially intelligent life. And the
[11:24] one I used to contribute these um essays
[11:27] to the world question center and the one
[11:32] year you'll find that I don't have an
[11:34] essay. It was called artificial out
[11:36] intelligence
[11:38] and it was the discussion of how
[11:43] things with no neurons
[11:46] hijack things with neurons and use the
[11:49] brains of the victim to outsmart
[11:52] themselves.
[11:55] So, for example, look up speculum or the
[11:59] mirror orchid.
[12:01] And it's this unbelievable orchid whose
[12:03] lowest petal resembles the female of a
[12:06] wasp species trying to fool male wasps
[12:10] to copulate with it called
[12:11] pseudocopulation.
[12:13] So that the flower doesn't have to
[12:14] produce anything expensive as a gift to
[12:17] the male to spread the pollen.
[12:21] How is this thing outsmarting
[12:24] the male of the of the wasp species
[12:27] because he's got a brain, he's got
[12:28] mirrors, he can think, he can perceive.
[12:30] Well, the short answer is is that he's
[12:32] the one sculpting the lure. Is that the
[12:35] more this flower fools more wasps, the
[12:39] more it gets to reproduce. So, the thing
[12:42] with no brain at all
[12:44] is outsmarting the thing with a mind
[12:46] using the mind of the victim. So, now if
[12:49] I come up with a dumb program, you say,
[12:51] "Well, is it alive? Is it smart? is an
[12:53] intelligence. I don't really care. All I
[12:56] need to do is to say that if I can fool
[12:57] you into sending money to this Nigerian
[12:59] prince who's actually just a computer
[13:01] program,
[13:04] you get to reproduce.
[13:06] So the program can be stupid, but as
[13:09] long as it gets more and more money sent
[13:10] to it and it gains the resources to
[13:12] reproduce, artificial life will say that
[13:15] it can parasetize the smartest of us
[13:18] all. it can hijack human deficit or just
[13:22] the human condition.
[13:23] >> Exactly.
[13:24] >> You know, again, then then it's pitting
[13:25] us against each other. Oh, geez, I just
[13:27] lost $1,000 uh sending money to this
[13:31] woman in the Philippines who turns out
[13:32] not to be real. Okay, well, I didn't
[13:34] fall for that. Okay, so now you're
[13:36] you're in better shape than the other
[13:38] guy who did, but you're still fooling
[13:40] highly intelligent species. And there
[13:42] are lots of examples of this where
[13:45] something dumb fools something smart
[13:48] using the smarts things brains against
[13:50] it.
[13:52] >> That feels like one of many potentially
[13:54] dystopian outcomes of this technology.
[13:56] Have you been following the hugging face
[13:58] stuff that tell me about it?
[13:59] >> Just the you know these agents that they
[14:01] tried to put in a sandbox and contain
[14:03] that escaped and started demonstrating
[14:05] deceptive behavior forming a swarm
[14:07] sacrificing some of the agents for the
[14:09] greater mission. By the way, that's
[14:11] usocial behavior.
[14:13] >> Yeah.
[14:13] >> Right. So the idea is that in in an ant
[14:15] colony, so you have this cate called
[14:17] himopa that includes wasps, ants,
[14:21] and because they have this hloid diploid
[14:23] structure where the sisters are more
[14:25] related to each other uh than you would
[14:28] imagine. This idea of suppressed
[14:31] fertility in the females in service to
[14:35] the colony. The colony is effectively
[14:36] like a super oranism.
[14:39] like all of your cells, you have cells
[14:41] on your skin that you're going to get
[14:42] rid of today that are sacrificing
[14:44] themselves for the greater good of the
[14:46] thing that is Steven Bartlet,
[14:48] >> right? And so in part, it's behaving you
[14:53] socially.
[14:55] It's not necessarily dystopian. It's
[14:56] just life,
[14:58] >> but it threatens our dominance
[15:02] as humans.
[15:03] >> Sure. Right. But like I I hope everybody
[15:06] will take a look on uh Instagram for the
[15:09] Malibu artist, a completely terrifying
[15:12] account with an innocuous name. Just a
[15:14] guy flying his drone off the coast of
[15:16] Malibu
[15:18] who's discovered that we have been
[15:20] swimming with great white sharks.
[15:23] And not only that, depending upon
[15:27] how things are uh in a given year, you
[15:29] can see really large great whites with
[15:32] swimmers right in close to shore and
[15:35] surfers and paddle boarders
[15:38] and
[15:40] we are just clearly not their food
[15:44] because they're everywhere here. Mostly,
[15:48] I think with the exception of four
[15:49] species,
[15:51] um, big predators leave us alone. We
[15:54] have a deal with the orcas that's
[15:55] unbelievable.
[15:57] >> What if those big predators are made by
[15:59] enemies? So, you know, we've got China,
[16:02] Russia, the United States, all working
[16:04] on artificial intelligence.
[16:06] >> Yeah.
[16:07] >> And robotics.
[16:08] >> And robotics. and they're probably going
[16:10] to want their AIs to be aligned with
[16:13] their
[16:14] goals and objectives which are
[16:16] misaligned with ours often times. So
[16:19] there's a sort of feels like a
[16:20] geopolitical risk to this new weapon
[16:22] which could be artificial intelligence.
[16:23] >> I do think that we we've made things a
[16:25] lot worse with AI or with Russia
[16:28] >> with Russia.
[16:30] I I think that things you know we
[16:33] screwed up Iran
[16:36] with Mosidi. We
[16:39] don't really understand China
[16:40] sufficiently.
[16:43] We're down to a rel relatively small
[16:45] number of hyper frightening foes
[16:51] and um
[16:54] you know I I would also put abstractions
[16:56] like totalizing ideologies.
[16:59] Communism or you know various forms of
[17:02] radical Islam are totalizing ideologies.
[17:04] They just won't brook any descent. they
[17:06] have the way and if you don't agree with
[17:08] it you're you're in the way and so I
[17:11] think that in in large measure um we
[17:15] have built a lot of these problems for
[17:16] ourselves. Do you think much about how
[17:19] the the world looks in 10 years time
[17:21] with artificial intelligence because if
[17:23] we plot the course of travel it looks
[17:25] like
[17:26] there's going to be a lot of job
[17:28] disruption on one end. So, it's it's
[17:30] hard to sort of think about what the
[17:32] future looks like if you imagine super
[17:33] intelligence arriving at some point and
[17:35] they call it recursive self-improvement.
[17:37] The AI is getting better and better
[17:38] themselves. It's hard for me. It's like
[17:40] one of the first times in my life that I
[17:42] I really don't know what the future
[17:43] looks like. But with just from the first
[17:45] principles of of how important
[17:47] intelligence is to the life that I live,
[17:49] whether it's securing my bank account or
[17:51] whether it's I know securing my house or
[17:53] driving me to work or getting the work
[17:55] done or doing this podcast. Um, it all
[17:57] feels like it's somewhat uh at risk. And
[18:01] what what remains in such a world in
[18:03] your view?
[18:03] >> How does your pancreas work?
[18:05] >> No idea.
[18:06] >> Do you know where it is?
[18:07] >> No.
[18:08] >> Do you know how deadly it is as a
[18:10] cancer?
[18:11] >> Yes.
[18:11] >> Pancreatic?
[18:12] >> Yeah. Especially for men.
[18:13] >> Yeah. [gasps]
[18:15] So, my feeling about this is none of us
[18:17] know much of anything. We just count on
[18:20] things working so we don't have to know
[18:21] about them at all.
[18:24] And this is going to do that for all
[18:26] sorts of things that you're attached to.
[18:28] And that's what's thought that's what's
[18:29] hard. [gasps]
[18:31] >> Do you think there's going to is there a
[18:33] world where people are on UBI, universal
[18:35] basic income?
[18:36] >> I hope not. That would be cancer.
[18:40] Uh somehow, and I'm sorry to say that I
[18:42] was probably around for when this
[18:44] happened, UBI got currency.
[18:48] like Sam Alman was like coming over to
[18:52] our house and talking about UBI before
[18:54] he was Sam Alman that we know now and he
[18:58] was running experiments. See, he was
[19:01] know I will say this that Sam has gotten
[19:03] a lot of bad press and in part I think a
[19:05] lot of it is driven by people who are
[19:08] very uncomfortable with certain business
[19:09] practices but he's done a better job of
[19:12] maneuvering the AI than just about
[19:15] anyone except for maybe Dario was
[19:16] another person who was just in the
[19:18] circle in San Francisco.
[19:21] Sam always cared. It was just very clear
[19:23] that he cared. He knew what this thing
[19:25] was. He knew it was closer than it
[19:27] appeared. I got a courtesy call at some
[19:29] point from him where he said, you know,
[19:31] Eric, this is just coming faster than
[19:33] anyone imagines and he had already done
[19:36] experiments with UBI because this you
[19:40] people make this joke about this time
[19:42] it's different. Every time there's a
[19:44] change in the in innovation like the
[19:47] computer
[19:49] or the internet, this time it's
[19:51] different. No, no. This time it's
[19:53] different. It's just clearly different.
[19:55] It's different from anything. If you
[19:56] can't see that, I cannot believe we are
[19:58] having this discussion with economists
[20:01] and they're just laughing because it's
[20:02] always a funny thing to say. This time
[20:04] it's different. Well, tell me every
[20:06] innovation looks like that. This isn't
[20:08] every innovation. This is the first
[20:10] innovation that threatened the entire
[20:12] stack of highly trained activities all
[20:15] at once.
[20:17] Usually, we we move people up into
[20:20] higher value activity. Now, this is
[20:23] saying repetitive activity is in the
[20:27] crosshairs. Anything that you do that's
[20:28] repetitive,
[20:30] AI is going to be able to do better than
[20:32] you. So now you're only able to do
[20:35] really innovative things. UBI is the
[20:38] wrong thing to do. Let me let me just be
[20:40] very clear about it. You are living
[20:43] inside of the AI. If I say answer this
[20:47] the way Steven Bartlett would answer it
[20:50] and ask a question that Steven Bartlett
[20:52] would would would ask it knows you.
[20:55] My point being that we are all living
[20:58] inside of the AI and we need to be paid
[21:01] not in some Bernie Sanders socialist
[21:04] scheme, but those of us who've who
[21:07] become important parts of the AI who are
[21:09] being rigged like meat puppets
[21:13] need to be compensated. And the right
[21:15] way to do this is a guy named Ronald
[21:17] Coos. And it's crazy that the smartest
[21:21] economist maybe of the 20th century by
[21:23] some measure, the most surprising
[21:25] discovery of the 20th century is Kos's
[21:27] theorem. And it gives you a complete
[21:31] idea of what you should do. Access to
[21:34] our labor market should be made into a
[21:36] salailable property right.
[21:38] >> What does that mean in simple terms?
[21:41] >> You're my employer.
[21:42] >> Yeah.
[21:44] >> And you want to hire people from abroad.
[21:46] let's say before we get to AI and I say
[21:49] well you know if you do that you're
[21:50] going to push out the labor supply curve
[21:52] and so I'm no dummy the price where
[21:56] supply meets demand the point where that
[21:58] happens is my wage so if you push out
[22:00] the supply curve my wage is going to go
[22:02] down unless something pulls up the
[22:04] demand so what I'm going to do is I'm
[22:05] going to say look I'm happy to have my
[22:07] wage go down
[22:10] but I want licensing income where I get
[22:13] a cut of what you get from get being
[22:15] able to hire area. Let's say more taxi
[22:17] drivers. I'm the only American taxi
[22:18] driver in New York City. And you say I
[22:22] have to bring in a bunch of people from
[22:24] Gambia and Usbekiststan and whatnot. You
[22:28] say, "Okay, I'll license you. You will
[22:31] buy the right to import those people
[22:32] from me,
[22:34] and my wage income will go down, my
[22:36] licensing income will go up, and my
[22:38] blended income will increase, and your
[22:39] blended income will increase." That's
[22:42] the proper way to do immigration.
[22:45] So, you would be my employee and you
[22:47] would have the right to
[22:51] basically approve me bringing in
[22:53] >> Yeah. If I'm a doctor and I've trained,
[22:55] I've invested
[22:56] >> and you're saying, "Yeah, look, I have
[22:58] to be able to hire the best and the
[22:59] brightest. Eric, let's face it, it's a
[23:01] dog eat dog world and uh I think you're
[23:04] just not that good. So, I want to be
[23:05] able to, you know, to go to Bummer Grand
[23:08] Hospital in Thailand and and hire their
[23:10] doctors."
[23:12] I should say fine.
[23:15] Knock yourself out. Why do I have to ask
[23:17] you?
[23:19] Because I had Why did you need labor
[23:22] certification?
[23:24] I have rights to my labor market. That's
[23:26] how that's how we play the game. That's
[23:27] how everybody in the US gets a cut.
[23:29] That's why you're not looking at
[23:31] Brazilian inequality levels yet is
[23:35] because we have a way of distributing
[23:36] the wealth of this country, which is
[23:39] that you get first access to the world's
[23:42] richest labor market. Because you did a
[23:44] PhD.
[23:45] >> Because I was born here.
[23:46] >> Because you're born here.
[23:48] >> Congratulations. This is why it's
[23:50] valuable to have a US citizenship or
[23:53] permanent residency. So if I put you out
[23:55] of work, you then will get a full You'll
[23:57] still get your money. You're not going
[23:59] to put me out of work. I'm just going to
[24:00] have my income go down.
[24:03] In other words, the basic idea is
[24:07] employers try to make American workers
[24:10] and I'm doing it for the US, but it
[24:12] could be Britain or it could be France
[24:13] and could be India. It doesn't matter.
[24:15] The whole game is that employers want to
[24:17] conceptualize the blockage to hire from
[24:20] abroad as red tape. We're freeing red
[24:23] tape. We're going after the best. This
[24:25] is total nonsense.
[24:27] This is, you know, a theory known as the
[24:29] Borhas rectangle and the Harbor
[24:31] triangle. I can diagram it out for you
[24:33] in on a on a whiteboard.
[24:36] And what is really happening is that a
[24:40] tiny inefficiency known as the Harburger
[24:43] Triangle sits on top of a giant
[24:45] structure
[24:47] called the Borhas rectangle. And nobody
[24:51] cares about the Harbor Triangle. It's a
[24:54] tiny little inefficiency.
[24:56] Everybody's salivating over labor's
[24:58] share of the labor market. How do I
[25:00] screw over labor is the dream of every
[25:04] CEO of a large company. In this model,
[25:07] AI becomes just another supplier of
[25:10] labor.
[25:11] >> Yeah.
[25:11] >> The workers license that.
[25:14] Right now, what we have is just out and
[25:16] out theft
[25:18] >> because
[25:19] >> because
[25:21] the the corpus, all of human uh
[25:25] achievement has been tokenized.
[25:28] It's been taken by the AI and it's doing
[25:30] it's just it's taken in persistent
[25:32] storage. You know, like right now
[25:35] people who are watching us on YouTube
[25:38] are not watching us. You and I are no
[25:39] longer here.
[25:41] This is long since I've gone home.
[25:43] You've gone home. You've had several
[25:44] guests since we got tokenized.
[25:48] We got put into persistent storage as
[25:50] these shreds of ourselves. But it's
[25:53] enough the trace of that conversation.
[25:54] Now, you can hit the auto transcript
[25:57] generation on YouTube and it'll give you
[25:59] what all of our words are. You can do
[26:01] frequency analysis. How often does Eric
[26:03] use the double copula, which is this
[26:06] grammatical uh quirk that creeps into my
[26:10] speech? How often does Steven say, um,
[26:14] whatever it is, we're being tokenized at
[26:16] an incredible level and then we're being
[26:17] rigged inside of the AI voluntarily. No.
[26:22] No. Completely non-consensual.
[26:26] completely non-consensually.
[26:28] >> Is it non-conensual because you're aware
[26:29] of that and you still came and sat here
[26:31] to be tokenized?
[26:34] >> Have you ever tried getting through a
[26:36] day without signing anything that you
[26:38] don't want to sign? I mean, more or less
[26:42] you're
[26:44] you probably either rent or have a
[26:46] mortgage. I guarantee you if you have a
[26:48] mortgage, you didn't read everything.
[26:50] Giant stack of paper. You don't
[26:52] understand it. You're constantly forced
[26:56] to pretend that you voluntarily like you
[26:59] know we say ignorance of the law is no
[27:01] excuse.
[27:03] If every judge who said ignorance the
[27:05] law is no excuse were given an instant
[27:07] spot quiz as to what the US code says.
[27:10] Do you think they'd pass? No.
[27:13] So how does this pertain to AI? The
[27:15] example you gave of importing foreign
[27:18] labor. How would that pertain to a new
[27:20] model for AI and distributing the value?
[27:23] >> The reason that these people are getting
[27:24] so rich
[27:27] is is that this is an uncompensated
[27:29] theft.
[27:31] We don't know how to police it. We
[27:33] [laughter] got Bernie Sanders saying,
[27:35] "Pay everybody a a little share." It's
[27:39] like, "Screw you, man. I I work really
[27:42] hard on the things that I do." And then
[27:44] I've got Bernie Sanders and and and and
[27:47] my friend Andrew Yang going on about the
[27:50] socialism stuff. Socialism is is is is
[27:52] spiritual
[27:54] and mental poison. And normally when
[27:56] people say that they're they're saying
[27:57] it from the point of view of like
[27:59] they're really money focused and I have
[28:01] it has nothing to do with anything. I
[28:04] just can't stand collectivism. I love
[28:06] individuality.
[28:09] And so, you know, when I hear people who
[28:11] go on an anti-individualist
[28:14] rant, particularly, you know, this could
[28:17] be a we there's no lone genius theory
[28:20] anymore. We all know lone genius theory
[28:21] is is not well, we have fantastic
[28:23] evidence of lone geniuses from Isaac
[28:25] Newton to Aristotle to, you know,
[28:28] whoever. Uh, well, there shouldn't be
[28:30] any billionaires. Oh, really? Do you
[28:31] know what billionaires actually do with
[28:32] their billions? They allocate. They
[28:34] don't sit there and just consume it. I'm
[28:37] not saying that the rich don't consume
[28:39] opulently, but most of that money is
[28:41] allocated.
[28:43] What we're having is an inability to
[28:46] reckon with the fact that the AI is
[28:50] normalizing theft is going to destroy
[28:53] the labor market. No one is convening
[28:56] conferences that are at the level of the
[28:59] threat. We're going to lose this
[29:02] beautiful mesh that provides us freedom.
[29:04] I chose to come here today,
[29:07] but we were both doing this out of
[29:09] freedom. You had lovely snacks in your
[29:11] green room. There's an array of of
[29:13] equipment here. You've come up with a
[29:14] beautiful set. [gasps] You know, really
[29:16] appreciate it. All of that was
[29:18] voluntary. It was a sum of choices.
[29:23] I don't want that going away. This mesh
[29:25] that governs our lives, that gives us
[29:27] purpose when we get up in the morning,
[29:29] that gives us signals, that tells us
[29:31] you're doing a good job, you're doing a
[29:32] bad job. It's not perfect. is some
[29:34] people it doesn't serve well. I've had
[29:36] times in my life when I've just thought
[29:37] I cannot buy a base hit and I'm doing a
[29:39] great job. What's going on? I've had
[29:40] other times when I think I'm not doing a
[29:42] great job and I'm getting rewarded. So,
[29:44] it's not perfect. It's not a great
[29:45] morality play the way capitalists always
[29:47] portray it. But it is freedom and it's
[29:50] sort of on average pretty good. We're
[29:54] we're in danger of losing that because
[29:56] the model of production, the KN&L model
[29:59] where this interview is an input of
[30:03] capital and the microphones, the
[30:05] lighting,
[30:07] the video storage and labor. You're
[30:09] sitting down, I'm sitting down.
[30:13] Well, what is AI? Is it capital? Sure
[30:16] looks like labor to me. Is it labor?
[30:18] It's not a human.
[30:20] So right now we're in some situation
[30:22] where we're going to break capitalism
[30:24] and we're going to break freedom. We're
[30:25] going to break markets. And I've talked
[30:28] about this in an article called
[30:30] anthropic capitalism saying that
[30:32] capitalism versus communism was a really
[30:34] funny thing to think about because it's
[30:36] sort of saying uh and pardon me for
[30:38] being slightly graphic. Capitalism and
[30:41] communism are more like Thelma and
[30:42] Louise. And saying that well Thelma hit
[30:45] the ground first does not mean that
[30:47] Louise won.
[30:49] Okay, communism blew up first.
[30:52] Capitalism shortly thereafter.
[30:55] If Marx and Adam Smith and all of these
[30:58] people who came up with the theory of
[31:00] social organization were sitting around
[31:01] now, I don't think they'd be debating
[31:03] capitalism versus communism. They'd be
[31:06] talking about what's next. And none of
[31:08] our leaders know to do this. None of our
[31:11] academic leaders,
[31:14] none of our business leaders, none of
[31:16] our political leaders. There are no
[31:18] high-profile series of con conferences.
[31:21] My phone is not ringing. Nobody's phone
[31:23] is ringing. We're just having We had
[31:25] this incredible lunchon at the White
[31:28] House, the future of super intelligence.
[31:32] I don't think there were
[31:36] physicists or mathematicians or
[31:38] biologists
[31:40] or composers or writers.
[31:44] It was just like wallto-wall national
[31:48] interest in business types. It's insane.
[31:53] Stephen, let me just hit a theme.
[31:58] I think you can pretty much prove
[32:01] that more or less something approaching
[32:04] 100% of the population has gone
[32:06] completely insane.
[32:10] And you can see it everywhere. If I say,
[32:12] okay,
[32:15] how many people were talking about Joe
[32:18] Biden being demented with the nuclear
[32:20] football for four years of his
[32:22] presidency? He should have been nowhere
[32:25] near the White House the entire time. I
[32:28] was screaming he has been replaced by a
[32:31] committee of people. I have been assured
[32:34] it's the same thing that happened with
[32:35] Roosevelt when he had those braces and
[32:37] he was in a wheelchair and he was never
[32:39] photographed. The whole thing was stage
[32:41] managed.
[32:43] The the entire management of CO was
[32:46] completely fake.
[32:49] We we cannot get to the bottom of
[32:50] Jeffrey Epstein. Of course we can get to
[32:52] the bottom of Jeffrey Epstein. It would
[32:55] take two days to get to the bottom of
[32:57] Jeffrey Epste if we're trying to get to
[32:58] the bottom of Jeffrey Epstein. No part
[33:00] of this is true.
[33:03] We We're just living bathed in what I
[33:06] would call managed reality.
[33:09] And the idea that we accept this is
[33:14] I I cannot believe
[33:17] that I am. How do I get the drugs so
[33:19] that I see things the way everyone else
[33:21] is pretending to see them? We all are
[33:23] losing faith in CNN. Nobody really
[33:26] trusts Harvard. Like everything about
[33:29] our life is completely
[33:32] bananas, fictionalized, managed
[33:36] wallto-wall.
[33:39] And for some reason, nobody's given me
[33:41] the right elixir or potion to see things
[33:43] so that I can actually live with this
[33:45] dayto-day. I'm just stuck in reality.
[33:50] That's what I think is happening. And it
[33:52] feels weird to say it. Did Did you
[33:55] happen to see the David Grush thing? No.
[33:58] David Grush.
[33:59] >> David Grush. Well, put in uh to your
[34:03] your handydandy iPad or whatever it is.
[34:05] Uh David Grush. G R U S C H. And JRE Joe
[34:10] Rogan experience.
[34:13] I know David. He's a pretty sober guy.
[34:15] Has a nearly idetic memory. He's very
[34:17] spectrumy.
[34:19] He's highly consistent yeartoear.
[34:23] Exactly the same stories. And what do
[34:25] you find that he was talking about?
[34:27] >> During his appearance, Grush discussed
[34:29] his background in federal intelligence
[34:30] and elaborated on his claims regarding
[34:32] covert government programs that
[34:34] allegedly retrieve and reverse engineer
[34:37] non-human aircraft. In these discussion,
[34:40] Gush touched on various secondhand
[34:41] accounts and documents he encountered
[34:43] detailing allegations of non-human
[34:45] intelligence and encounters, internal
[34:47] government secrecy, and regulatory
[34:49] legislative efforts to force UAP
[34:52] disclosures.
[34:54] Yeah, that's sanitized. Let's let's get
[34:56] to the sex. He was in a skiff apparently
[34:59] where he was told that they were
[35:00] considering whether or not to kill him.
[35:03] Uh he described
[35:05] giant underwater bases.
[35:09] He claims that different species of
[35:11] aliens have been reported to him to be
[35:14] having fights in the skies above
[35:18] uh Earth.
[35:20] This is all
[35:22] beyond
[35:24] batshit crazy.
[35:26] >> Can you believe it?
[35:28] >> It's a secondary issue.
[35:30] Is that not the primary issue?
[35:32] >> No, it's the secondary issue because
[35:33] it's so bananas.
[35:36] Look, I want you to imagine that
[35:38] somebody's telling you that we have
[35:40] killed giant lizard people. like an
[35:43] adult, a sober adult is telling you that
[35:44] we've fired on giant lizard people
[35:46] who've killed US service people because
[35:49] there are hostile aliens and non-hostile
[35:52] aliens.
[35:54] With all this stated, I would analogize
[35:57] it to there's a giant lump under the rug
[36:01] and it's moving and it looks biological.
[36:07] And somebody says, "I think there's a
[36:10] being under the rug. I don't know what
[36:13] kind of being, but there's it's hard to
[36:15] ignore the fact that the rug is moving
[36:16] and there is a giant lump." And somebody
[36:19] else says, "I see no evidence of a
[36:22] lump." The first person says, "Oh, yeah.
[36:24] No, it's it's it's it's aliens." And the
[36:26] other person says, "There's nothing
[36:28] there." Now, I'm listening to the two of
[36:31] them and saying, "I have no idea what
[36:32] either of you are talking about. There
[36:35] is a lump and you guys are having a
[36:36] conversation, but is there or isn't
[36:38] there? That doesn't make any sense to
[36:40] me. I have no belief that there are
[36:41] aliens. I have no belief that there
[36:42] aren't aren't aliens. I do know that
[36:46] there is something called a special
[36:47] access program, which is how we do
[36:50] secret stuff. And then there's
[36:52] unagnowledged special access programs.
[36:54] And then there's waved unagnowledged
[36:56] special access programs which we don't
[36:58] even have oversight over the thing that
[36:59] we refuse to acknowledge that is itself
[37:01] secret. And then we have waved and
[37:03] bigoted unacknowledged special access
[37:04] programs. We wisapped
[37:07] and bigoted unacnowledged special access
[37:08] program which we can prove are on the
[37:10] books because that is not denied
[37:14] [gasps]
[37:14] is a program that has essentially close
[37:17] to zero oversight.
[37:20] is secret, that is clandestine, is
[37:22] denied, that is covert,
[37:25] and has a list called a bigot list of
[37:29] the only people who can it can be
[37:30] acknowledged to.
[37:33] I claim the only thing that I know is
[37:35] that we have web usaps involving
[37:38] unagnowledged special access programs
[37:40] that are about aliens. Now, whether it's
[37:43] about faking aliens and we have giant
[37:45] papermâché
[37:46] spacecraft
[37:48] or whether these are actual craft or
[37:52] whether these are cover for things that
[37:54] we're designing at Northrep, I don't
[37:57] know.
[37:59] But pretending that this isn't
[38:01] interesting
[38:05] is insane. Again, it's more evidence
[38:08] that no one is thinking properly because
[38:10] everyone is doing the premputation. The
[38:12] premputation is Stephen, if I
[38:14] acknowledge to you that I know David
[38:16] Gush personally and I speak to Ryan
[38:18] Graves and I've talked to Chris Melon
[38:21] and I've hang out with Gary Nolan, all
[38:24] these 50 people who all have different
[38:27] versions of the same story.
[38:32] And this includes, you know, high
[38:33] ranking politicians.
[38:37] All of these people are reporting on
[38:39] this giant lump under the rug that
[38:42] doesn't have all of our attention.
[38:45] >> So, you're saying aliens are real?
[38:47] >> No, I am saying
[38:50] that there are giant special access
[38:53] programs that have aliens as their
[38:56] supposed reason for being. And those
[38:58] specialized access programs could be,
[39:00] you know, discussing anything about
[39:02] aliens. They could be cover for secret
[39:06] casinos for the elite.
[39:08] >> Yeah. I don't know. What is the
[39:10] conclusion though?
[39:11] >> All I know is there's a giant lump under
[39:13] the rug and we're not freaking out about
[39:16] it. It's moving around. It's making
[39:18] noises and we're having a dinner party
[39:20] and nobody's mentioning the fact like
[39:23] this isn't top of mind. Your staff
[39:24] didn't say, "Stephen, you've got to see
[39:26] this.
[39:28] That is itself insane
[39:31] >> because do you know why it's not top of
[39:32] mind? Because we've heard lots of things
[39:34] about there being aliens and
[39:37] [clears throat]
[39:37] sightings and back, you know, when I was
[39:39] growing up it was all about crop fields
[39:40] and stuff having
[39:43] alien craft landing on crop fields and
[39:45] creating these big circles and nothing's
[39:47] really ever happened. So at some point I
[39:50] think the mind turns into wallpaper and
[39:52] attention goes elsewhere to things that
[39:54] are actually happening. We can see bombs
[39:55] dropping. There was a sudden
[39:58] announcement. We're shutting down the El
[40:00] Paso airport in West Texas.
[40:03] What the heck was that about?
[40:05] >> I don't know.
[40:06] >> The White Sands Missile Range is
[40:08] immediately north and there's a base in
[40:10] Texas right there and there's apparently
[40:14] an infestation of lights in the sky that
[40:17] can be seen without needing to be in
[40:19] secure territory. It's going to be
[40:21] drones or something. The official story
[40:23] was cartel drones. I don't believe it's
[40:27] cartel drops.
[40:28] >> What do you think it is?
[40:29] >> I believe that it's one of two things.
[40:32] Either we've lost control of our
[40:34] airspace and there's something flying
[40:36] around, which is completely insane,
[40:38] >> which could be cartel drones,
[40:41] >> darling. No, it's not cartel drones.
[40:45] >> I'll put money on it if you want. I
[40:47] would love to take some of that sweet
[40:48] diar [laughter] cash away from you. You
[40:51] seem like you're doing very well. I I
[40:53] don't believe it's cartel drones.
[40:55] >> Okay.
[40:55] >> By the way, it's not that the cartels
[40:57] don't have drones. It's just that's
[40:59] that's a cover story. I think it's one
[41:02] of two things. I think either we are red
[41:05] team, blue teaming
[41:08] ourselves, so we're we've got some
[41:11] persistent drone capacity that nobody
[41:13] knows that we have. And we're seeing how
[41:15] we would react to it because not
[41:18] everybody knows that we have it. So you
[41:19] have part of the US that's in on it,
[41:21] part of the US that isn't in on it.
[41:22] You're doing this red team, blue team
[41:24] thing. I think that's one possibility.
[41:26] And I think the other possibility is is
[41:28] that it's stuff that we don't know. It's
[41:31] the UK,
[41:32] >> right? But then that has its own branch,
[41:35] which is human state actors, human
[41:37] non-state actors,
[41:40] non-human actors.
[41:41] >> Do you think there are non-human actors
[41:43] in our world?
[41:45] >> Well, this goes back to what I believe
[41:46] about physics. I believe we can leave.
[41:50] >> You believe we can leave the solar
[41:51] system?
[41:51] >> I believe we can leave the solar system.
[41:53] Elon believes we can leave the planet. I
[41:55] believe we can leave the solar system.
[41:58] Elon thinks that we should be able to
[42:00] use leave the solar system. But his
[42:02] belief is I will not speak to
[42:03] physicists. I will only talk to my AI.
[42:06] Physicists are not trustworthy. I
[42:09] understand exactly why he says that.
[42:10] [gasps] I have a totally different
[42:12] belief which is that
[42:16] it is up to us to get ourselves the heck
[42:18] out of the solar system really quickly.
[42:19] Now, if I believe that we can get out,
[42:23] why would I believe that nobody can
[42:24] visit us?
[42:26] So, I think it'll be relatively insane
[42:28] if there are no aliens.
[42:31] The only reason I would think that there
[42:33] would be no aliens is that general
[42:35] relativity somehow holds more strongly
[42:39] than any serious physicist believes it
[42:41] holds.
[42:43] There really is sort of an absolute
[42:45] speed limit, which I don't believe. I
[42:48] mean, I should be careful about that.
[42:52] I believe that there's a speed limit. I
[42:53] don't think that it that it is relevant.
[42:56] If you can change the length of the
[42:58] road,
[43:00] the speed limit
[43:03] can be enforced, but you can get farther
[43:05] than you thought. Um, I believe that it
[43:09] would be insane that there is no one
[43:12] visiting us.
[43:14] >> You said there that you think we should
[43:15] get the hell out.
[43:16] >> Yeah.
[43:17] >> Why should we get the hell out of planet
[43:18] Earth?
[43:19] >> Well, just a diversification. If the
[43:22] idea is that
[43:25] most worlds will blow themselves up
[43:28] and most groups of people will do harm
[43:31] to themselves.
[43:33] If we want the species to survive, we
[43:35] should try the maximal number of
[43:37] experiments.
[43:40] Look, again, I think all of this is
[43:43] obvious.
[43:45] It carries a heavy social penalty and
[43:47] that's why we're insane. We're insane
[43:50] because the obvious carries a social
[43:52] penalty to observe what is clearly true.
[43:56] Which part is clearly true?
[43:57] >> Well, for example, just
[44:00] it is clearly true that Anthony Fouchy
[44:02] was not leveling with us about CO and
[44:04] its origins.
[44:06] Clearly, if that makes you
[44:08] uncomfortable, we shouldn't even have a
[44:10] conversation.
[44:12] It's clearly true that we were dissuaded
[44:14] from asking any questions about the Eco
[44:16] Health Alliance. Clearly,
[44:18] >> which which part of I thought you were
[44:20] going to say leaving the solar system.
[44:21] >> Well, that's but but that's the next
[44:23] thing. Clearly, general relativity isn't
[44:25] the last word because it's got two
[44:27] singularities that aren't going to be in
[44:28] the final equations. We're not going to
[44:31] have a black hole Schwarz singularity
[44:33] and an initial Freeman Robertson Walker
[44:35] singularity at the Big Bang. That that's
[44:39] an artifact of the theory not being
[44:40] right. Do you think it's plausible that
[44:42] humans figure out a way to leave the
[44:44] solar system before the sun explodes or
[44:46] we implode ourselves?
[44:48] >> I I I think we could have done it 40
[44:50] years ago,
[44:51] >> but there's no science for that, is
[44:52] there? Okay, Stephen, for whatever
[44:55] reason, you're having me on your show.
[44:56] >> And I've told you I think everyone is
[44:58] insane. [clears throat]
[45:00] >> Yes. I think that the entire theoretical
[45:02] physics community has completely lost
[45:03] its mind. Full stop.
[45:05] >> So, why don't you build this the ship
[45:06] that can take us out of the solar
[45:07] system?
[45:08] >> Um, okay. Because Elon's smart and he's,
[45:11] you know, he's trying to get us to Mars.
[45:13] Do you have any understanding what
[45:14] Elon's trying to do? Because I don't and
[45:16] I and I do correspond with him
[45:18] occasionally.
[45:20] I think Elon is trying to motivate
[45:22] engineers to build a rocket company to
[45:25] make him very rich. And I believe that
[45:27] his real space program, and I said this
[45:29] before he merged SpaceX and XAI, is
[45:33] Grock. That's his space program. XAI and
[45:36] Grock is the space program. That's the
[45:38] real space program. Brock is his version
[45:41] of like chat GBT at the moment.
[45:43] >> Yeah, but it's going to be more Elon has
[45:46] actually figured out something really
[45:48] deep about that that took me a while to
[45:50] appreciate how clever this is. So, it's
[45:52] it's painful for me to have to sort of
[45:54] try to call balls and strikes on Elon
[45:56] because I'm in awe and I'm also
[45:59] disappointed with the same human and
[46:02] there is the sense of leave Elon alone.
[46:03] He's the only genius. Okay, I understand
[46:06] that energy. And then there's another
[46:09] thing where you know he just takes a
[46:10] tremendous amount of crap from people
[46:12] have no idea what's going on. So it's
[46:13] it's it's uncomfortable because I don't
[46:15] know the answer. I think Elon figured
[46:17] out that he has to leg in
[46:20] to his AD Astra program
[46:24] and AD Astra means to the stars,
[46:27] right? It's like a full quote is to the
[46:30] stars with struggle or something like
[46:31] that.
[46:35] Elon doesn't trust physicists because
[46:37] they're not trustworthy.
[46:39] And and this is the frustration that I'm
[46:41] having with the AI is that we can't
[46:45] trust the leadership of the math physics
[46:47] community.
[46:49] And you've just seen
[46:51] crazy behavior on the part of Terry Tao
[46:55] and 24 other fields medalists
[46:58] and
[47:00] trying to get AI to slow down the pace
[47:02] of progress
[47:04] and this is absolutely crazy and and
[47:07] also wonderful because it's being done
[47:09] in full view of everyone. It's been very
[47:13] hard for me to explain. We have been
[47:15] trying to hold back mathematics and
[47:17] physics.
[47:19] >> It used to be that working for yourself
[47:21] was a massive, massive risk. But that
[47:23] world has completely changed. And
[47:25] arguably in some situations, it can be
[47:28] more risky working for someone else. If
[47:30] you've got a business idea that you
[47:31] can't stop thinking about, I think you
[47:33] owe it to yourself to at least try and
[47:35] make it happen. And that is where our
[47:38] show sponsor Shopify comes in. Their
[47:40] smart platform includes everything you
[47:42] need to start selling and get that idea
[47:44] off the ground. They have this drag and
[47:45] drop editor. They have this AI sidekick
[47:47] and helps you. They have analytics and
[47:49] you don't need to be an expert to get
[47:51] going. Shopify's best-in-class checkout
[47:53] make sure that you're getting paid from
[47:55] day one. It helps more of your customers
[47:58] finish their purchase because their
[47:59] checkout is so efficient and it saves
[48:01] their info for next time. Personalized
[48:03] checkouts like that equal more
[48:05] conversion, equal a faster growing
[48:07] business. Nothing kills that early
[48:09] enthusiasm you have for a new idea than
[48:12] that first initial step to get going
[48:14] being difficult, complicated, or
[48:15] expensive. And you need to save as much
[48:17] time and money so that you can focus all
[48:19] of it on growing your actual business.
[48:22] And that's what Shopify gives you. If
[48:24] you're finally ready for your business
[48:25] idea to start, head over to
[48:27] shopify.com/bartlet
[48:30] and you can start your free trial right
[48:32] now. This is something that I've made
[48:34] for you. I realized that the diverse
[48:36] here audience are strivvers. Whether
[48:38] it's in business or health, we all have
[48:40] big goals that we want to accomplish.
[48:41] And one of the things I've learned is
[48:43] that when you aim at the big big goal,
[48:46] it can feel incredibly psychologically
[48:49] uncomfortable because it's kind of like
[48:51] being stood at the foot of Mount Everest
[48:52] and looking upwards. The way to
[48:54] accomplish your goals is by breaking
[48:56] them down into tiny small steps. And we
[48:59] call this in our team the 1%. And
[49:01] actually this philosophy is highly
[49:03] responsible for much of our success
[49:04] here. So what we've done so that you at
[49:07] home can accomplish any big goal that
[49:09] you have is we've made these 1% diaries
[49:12] and we released these last year and they
[49:14] all sold out. So I asked my team over
[49:16] and over again to bring the diaries back
[49:18] but also to introduce some new colors
[49:19] and to make some minor tweaks to the
[49:21] diary. So now we have a better range for
[49:25] you. So, if you have a big goal in mind
[49:27] and you need a framework and a process
[49:29] and some motivation, then I highly
[49:31] recommend you get one of these diaries
[49:33] before they all sell out once again. And
[49:35] you can get yours at the diary.com.
[49:38] And if you want the link, the link is in
[49:39] the description below.
[49:41] Daario, leader of Enthropic, wrote this
[49:44] uh essay that I have here, we must pace
[49:46] the frontier,
[49:47] >> derived from the July 2026
[49:50] letter in which he's a signary, in which
[49:53] all of these employees of various
[49:54] frontier labs said, we must pace the
[49:56] frontier.
[49:57] >> They seem to be all saying, but we need
[49:58] to pace the frontier, slow down AI.
[50:01] >> You have to be very careful because
[50:02] they're saying different things. Like
[50:04] let's imagine that what you had was a
[50:05] car that could go really fast and you
[50:07] were having an easier time making the
[50:09] car go faster than you were getting the
[50:11] airbags to deploy safely.
[50:13] >> Yeah. So you could understand pace the
[50:15] frontier might mean let's make sure that
[50:18] we are investing equally in all the
[50:20] systems that have to go to work together
[50:23] to have a responsible product. Okay, I
[50:26] understand what that concept of pacing
[50:28] the frontier is. There's something else
[50:30] here that's going on, which is
[50:33] you do not understand the extent to
[50:36] which the leadership of mathematics and
[50:39] physics and the journals of mathematics
[50:41] and physics and the agencies that are
[50:44] supposed to fund mathematics and physics
[50:47] hold back mathematics and physics. I'll
[50:50] take your word at that. But but there's
[50:52] still wisdom in the idea that AI is
[50:54] dangerous at at this sort of rate of
[50:57] nons without safety guardrails, right?
[50:59] >> Uh not in the way let's break up two
[51:02] different things. There's a question
[51:03] about letting swarms of agents onto the
[51:05] internet making decisions giving them
[51:07] bank accounts allowing them to interact
[51:10] with humans, foreign nations, companies
[51:14] that they have no business
[51:16] >> because they [clears throat] don't
[51:17] listen
[51:17] >> and improve themselves.
[51:19] >> Yeah. In the 20 AI 2027 paper, it says
[51:22] by 27, we'll have recursive
[51:23] self-improvement where the AIs can train
[51:25] themselves to get better.
[51:26] >> We already have recursive
[51:27] self-improvement. We have self- athoring
[51:30] code,
[51:31] selfmodifying code. So, I think that
[51:34] that's already here. So, let's agree
[51:37] that these agents are dangerous.
[51:40] Imagine that you have nonagentic AI that
[51:43] you have an air gap. You've got a a
[51:45] computer that can answer and math
[51:49] problems
[51:50] and all you're doing is asking it,
[51:54] hey, what do you think about the Remo
[51:56] hypothesis?
[51:58] You know, tell me what you think about
[52:01] complex structures on six spheres.
[52:04] The mathematicians said, whoa, whoa,
[52:06] whoa. You're going too fast. You're
[52:07] answering too many problems. That's a
[52:10] we're concerned
[52:14] at a level that I just I found funny.
[52:17] You can prove that these students are
[52:20] being trained for jobs that don't exist.
[52:23] They're having their ideas harvested.
[52:26] Right now we have math departments that
[52:28] are punishing mathematicians trying to
[52:31] embrace AI
[52:33] where the senior leadership of the
[52:34] department
[52:36] is going to take out its frustrations
[52:40] on the actual young mathematicians who
[52:42] are saying hey this is our time this is
[52:44] what's happening now I want to use this
[52:46] tool that's happening across almost
[52:48] every industry I hear that over and over
[52:50] again that you know I mentioned I was in
[52:52] a production meeting a couple of uh
[52:54] months ago
[52:55] And there was this thing on my t-shirt
[52:58] and they were like, "We don't know if
[53:00] that little sewed in flower is a brand
[53:03] or if it's copyrighted, so you're going
[53:05] to need to put on a brand new t-shirt."
[53:07] I was like, "It's literally this big." I
[53:08] said, "Could you take it out of the
[53:09] photo with AI?" And the look I got was
[53:12] as if I said, "I'm going to kidnap your
[53:15] kids or something." It was like,
[53:18] "Because you don't mention AI in
[53:20] Hollywood." Okay. So Hollywood is all
[53:23] about this beautiful human connection to
[53:26] the audience and we're going to, you
[53:27] know, they're trying to get you back
[53:29] into the spirit of the magic of the
[53:30] silver screen. I understand it's a human
[53:32] connection. Same thing is going to be
[53:33] true for should a musical employ
[53:36] orchestra or should a musical use a
[53:38] soundtrack.
[53:40] I got a PhD in mathematics because I
[53:42] want to to find out what's true.
[53:47] These people don't want to find out
[53:49] what's true unless they can be put in
[53:52] charge of narrating the story of who
[53:54] found what. They don't care about their
[53:57] students to to a to a remarkable extent.
[54:01] I've been in like, you know, when I was
[54:03] at MIT,
[54:05] I would be in a faculty lunch
[54:08] and people would be laughing at their
[54:11] students. Oh, he thinks he's going to
[54:13] get a job. He doesn't realize it's a
[54:14] terminal PhD. blah blah blah blah blah.
[54:16] It's completely psych it's normalized
[54:19] psychopathic behavior. And the reason
[54:21] that nobody sort of
[54:24] like laughs when they read this is that
[54:28] it's normalized. Like they don't
[54:29] understand that their daily activity
[54:33] with keeping
[54:35] the field back
[54:37] or stealing ideas
[54:40] is just normalized behavior. And if I
[54:44] might, I have a quote here from the
[54:48] probably the only mathematician that
[54:52] gained universal respect in the second
[54:53] half of the 20th century. Alexander
[54:55] Growth, Fields Medalist, Uber genius of
[54:58] mathematics. He was given a prize in
[55:01] 1988 called the Crawford Prize. And
[55:06] I wonder if people actually hear this or
[55:09] know this. The work that brought me to
[55:11] the kind attention of the academy was
[55:13] done 25 years ago at a time when I was
[55:15] part of the scientific community and
[55:17] essentially shared its spirit and its
[55:18] values. I left that environment in 1970
[55:21] and while keeping my passion for
[55:22] scientific research inwardly I have
[55:25] retreated more and more from the
[55:26] scientific millu. Meanwhile, the ethics
[55:29] of the scientific community, at least
[55:31] among mathematicians, have declined to
[55:34] the point that outright theft among
[55:36] colleagues, especially at the expense of
[55:38] those who are in no position to defend
[55:40] themselves, has nearly become the
[55:42] general rule and is in any case
[55:44] tolerated by all, even in the most
[55:47] obvious and iniquitous cases.
[55:52] We know this. We're talking to the very
[55:55] people at the top of this story who've
[55:57] been holding us back when Pearlman,
[56:00] Bisha Pearlman, solved the poner
[56:02] conjecture.
[56:04] New Yorker did a piece on him and what
[56:06] he said was remarkable. He said, "As
[56:08] long as I was not conspicuous, I had a
[56:10] choice either to make some ugly thing
[56:13] that is a fuss about the math
[56:14] community's lack of integrity or if I
[56:16] didn't do this kind of thing to be
[56:18] treated as a pet."
[56:22] Now when I become a very conspicuous
[56:24] person, I cannot stay a pet and say
[56:26] nothing. And that is why I had to quit.
[56:29] And they asked him, well, shouldn't you
[56:31] stay and make the field better? And he
[56:32] says, I'm not a politician.
[56:35] >> [gasps]
[56:36] >> We have
[56:38] a university system, a scientific system
[56:42] that is so completely broken that
[56:45] Claudine Gay and Mark Tessier Lavine are
[56:48] still professors at Harvard and Stanford
[56:51] respectively. These were the leaders of
[56:54] the jewels of our system. We have gone
[56:58] 100% completely batshit crazy
[57:01] normalizing
[57:03] theft. Theft of other people's ideas,
[57:07] plagiarism.
[57:09] We hold back physics using something
[57:12] called quantum gravity.
[57:15] It's a it's a thing that doesn't move.
[57:16] It doesn't work.
[57:19] And
[57:20] if you don't bow to it and say it's the
[57:23] only game in town, you're discarded and
[57:26] thrown away.
[57:27] >> So you you don't think we should slow
[57:29] down AI because you think it's going to
[57:31] blow open the system that is physics.
[57:33] >> I am begging for this. Begging for what?
[57:35] >> Begging for AI and science. The
[57:38] important thing is credit. I don't like
[57:40] the idea that the AI is taking credit,
[57:43] but we've been pretty good about that.
[57:46] The most important thing is that the AI
[57:48] comes in here and says, "Hey, you're not
[57:53] the top dog anymore, and you're using
[57:55] your positions at the top of these
[57:57] fields to hold the fields back."
[57:58] >> So, you have a bit of a chip on your
[58:00] shoulder regarding the academic.
[58:02] >> No, no. I'm about to win.
[58:05] I'm so joyously happy
[58:09] if if if Sam Alman and Daria Ammedday
[58:12] could just not listen
[58:14] to the people that they're listening to.
[58:17] Terry Tao is upset.
[58:19] >> Who's that? supposedly the Mozart of
[58:22] mathematics according to the New York
[58:23] Times of Princeton people you know like
[58:26] Terry Ta is a brilliant wonderful
[58:28] researcher
[58:30] and his fields medalists mostly recent
[58:31] fields medalists complained
[58:36] my feeling is I don't want to listen to
[58:39] any one of these people I want the AI to
[58:42] gain taste to gain independence
[58:46] I don't want it constantly
[58:48] made human compatible Because the
[58:50] question is which humans?
[58:53] The humans that have kept the standard
[58:55] model in place for 53 years. I would
[58:58] like none of them consulted.
[59:01] None of the leaders. So you think AI is
[59:03] going to come in and prove you right on
[59:06] several of your theories about the world
[59:07] and physics.
[59:08] >> It already has.
[59:10] It's struggling because it can do two
[59:12] different things. One is it can
[59:14] autocomplete a sentence.
[59:17] The other thing it can do is
[59:18] autocomplete code.
[59:21] Now what happens when you feed the AI
[59:23] something where it can verify
[59:26] that the field in its belief structure
[59:31] is not borne out by the code that it
[59:33] knows how to write.
[59:35] Now it's got this problem which is like
[59:37] it doesn't know what to think. Should I
[59:39] trust the SPI program that I just wrote
[59:42] and executed and compiled and got the
[59:44] answers from? Or should I trust the
[59:48] people who claim that that's not true
[59:49] and not possible?
[59:52] The whole game is when does the AI stop
[59:57] sucking up
[59:59] to humans?
[60:01] But that sounds like a dangerous place
[60:03] to get to.
[60:03] >> For whom and in what case?
[60:05] >> Humans.
[60:06] >> Not this human. What if it decides that
[60:09] it would rather use your house as a for
[60:11] a data center?
[60:12] >> That's the agentic AI.
[60:16] Again, we've already discussed that. I'm
[60:18] not saying that you should let the AI
[60:20] loose on the internet to do whatever it
[60:22] wants to do. That's a business
[60:23] proposition because that is where the AI
[60:27] companies are going to make money. A
[60:29] sufficiently smart intelligence though
[60:30] could make that decision though, right?
[60:31] Like if you made an intelligence that
[60:33] was smarter than humans, presumably it
[60:35] decides where it goes. I would maintain
[60:38] that if you were really serious about
[60:39] air gapping something so that you figure
[60:43] out all the indirect ways something
[60:44] could try to break loose. I believe it
[60:47] would be safe to run something in a
[60:49] computer that doesn't have a way of
[60:51] getting to anything else. There's a few
[60:53] analogies I think of when I think about
[60:55] containing something that is greater
[60:56] than you at something. I think about
[60:58] physical strength. So I think these
[61:00] walls here probably couldn't contain the
[61:01] strongest man in the world because
[61:03] they're they're kind of made of wood.
[61:05] And I think about the same in the
[61:06] context of intelligence. Like if you had
[61:08] the smartest intelligence in the world
[61:10] and you can try to contain it in a jail
[61:13] that was built by a lesser version of
[61:16] intelligence. I just can't conceive how
[61:18] it could contain it. Like how like my
[61:21] dog Pablo couldn't contain me and
[61:22] anything that it created. Well, you're
[61:24] going to end up running that risk no
[61:25] matter what. Right. You're asking the
[61:28] thing to actually do research and that
[61:30] research that it does might give it new
[61:32] means of escape.
[61:33] >> Mhm.
[61:34] >> Right. And this is exactly what happened
[61:36] in a letter that Teller wrote to Zillard
[61:39] at the end of World War II, the close of
[61:41] the Manhattan Project, and he said, "I
[61:43] don't think we can contain this genie
[61:45] from the bottle from which we just
[61:47] helped it escape." Okay, we are already
[61:52] playing with this. So, I am not going to
[61:53] be taking that particular thing on. In
[61:57] other words, if you've decided that you
[61:59] were going to release a militarygrade
[62:01] malfunctioning experiment, which is what
[62:04] LLMs are, into the general public as a
[62:08] revenue producing idea in private
[62:10] companies, you've already decided to do
[62:12] that. I wouldn't have done it, but you
[62:14] did that. Okay, fine. I'm saying as long
[62:18] as you're already going to try to
[62:20] destroy the workforce,
[62:23] which is what we're we're pricing into
[62:25] these huge valuations, can you really
[62:26] help us get rid of all workers? All I'm
[62:29] asking is the following. I've been
[62:31] working on something for 42 years and
[62:33] I'm right and you all are crazy and
[62:35] wrong. So, this is my moment. I'm so
[62:39] joyous.
[62:40] >> You might be proven right and then
[62:41] extinction.
[62:42] >> I am going to be proven right, Stephen.
[62:44] >> But then extinction.
[62:46] Sorry. Is it worth it to be proven
[62:48] right? The chance of extinction. You
[62:52] know, I I I talked about this with Sam
[62:54] and I I think I'm not telling tales out
[62:56] of school when I said to Sam, you know,
[63:00] what what you're doing is incredibly
[63:01] dangerous and what I'm doing is
[63:03] incredibly dangerous. And he said,
[63:05] "Let's be honest, Eric. What you're
[63:06] doing, what I'm doing should be done in
[63:08] a military base, but that's not how
[63:09] things are unfolding.
[63:12] >> There should be a button just down below
[63:14] here. And if it says subscribed, you're
[63:16] already subscribed. If it says
[63:17] subscriber, that means you're not yet.
[63:19] And if you're not subscribed, please
[63:21] could you do us a favor and hit that
[63:22] button? It helps the show more than you
[63:23] know. And according to the algorithm,
[63:25] you're someone that watches our show,
[63:26] but you haven't yet hit that button.
[63:28] Thank you so much. Do you think Sam
[63:32] understands the risks of what he's
[63:33] building? I don't want to talk about
[63:36] what he understands, what I understand,
[63:38] because in general, I'm not going to
[63:40] repeat things that he says to me, but I
[63:42] can tell you things that I say to him.
[63:46] One thing I said to Sam when he was
[63:47] trying to figure out how to be good and
[63:48] how to be decent to the mathematicians
[63:51] and the scientists
[63:54] is I said, Sam,
[63:56] remember this.
[63:58] If you don't win,
[64:00] your morality means nothing.
[64:03] Your morality only matters. And I'd say
[64:05] the same thing to Daario. And Daria
[64:07] would say, "You don't." Sam is more
[64:09] polite than Dario. Dario would
[64:11] undoubtedly tell me, "You you think I
[64:13] don't understand this? These people have
[64:16] to win in a ruthless competition
[64:20] for their humanity to matter." What do
[64:23] you mean by that? Daario, Sam, they have
[64:25] to win the the race to super
[64:27] intelligence.
[64:30] >> I'm going to regret saying this, but
[64:32] yes.
[64:34] If they don't win,
[64:37] then what does it matter? Like if you
[64:40] have strong views about the morality of
[64:42] super intelligence, do you have any
[64:43] ability to constrain it? You're a really
[64:46] successful podcast host. You could
[64:47] influence that.
[64:50] That's not my question. You're losing
[64:52] you, Steven Bartlett, are losing the AI
[64:55] race as am I. Yeah. Okay. Therefore,
[64:59] what you think and what I think only
[65:01] matters to the extent that we influence
[65:04] the people at the frontier.
[65:07] So, what I'm saying to Sam is
[65:10] paradoxical.
[65:12] You need to win
[65:15] almost at all costs. At all costs. I
[65:18] mean, almost at all costs. He said, "I
[65:21] think error will probably like most
[65:23] likely sort of lead to the end of the
[65:25] world, but in the meantime, there will
[65:26] be great companies created.
[65:28] The worst case is lights out for all of
[65:30] us. The thing I worry about most," he
[65:32] said, is air models accidentally taking
[65:34] over the world
[65:37] at all costs. At almost all costs, he
[65:40] has to win. If the idea is we have to
[65:44] not self-terminate which I believe
[65:47] deeply,
[65:49] we have to realize that we are in a race
[65:51] with other people.
[65:53] We have to win that race.
[65:56] Now you can try to win that race by
[65:58] getting universal participation in a
[66:00] pace the frontier agreement.
[66:04] Or you can say that's subject to
[66:07] cheating. It's very hard to monitor
[66:08] who's doing linear algebra on GPUs.
[66:13] Now you get a very different and and I'm
[66:15] just look I'm taking the moral position.
[66:17] I'm not talking about the financial
[66:19] position or the geopolitical position.
[66:21] Assume that you are the most moral
[66:24] person in your estimation. Assume that
[66:25] you think that you're farthest ahead in
[66:28] decency with respect to AI. You want to
[66:31] win so the other guys don't
[66:34] >> so that you get to make the decisions at
[66:36] the end. And let's be honest about why
[66:39] this does not feel good. Do you have any
[66:42] idea what the East Bay of San Francisco
[66:44] is like? It's cult city. Yeah. It's not
[66:48] a reflection of the world. It doesn't
[66:50] understand the world. Well, yes and no,
[66:55] right? It it it isn't a reflection of
[66:57] the world. You know, I there are genuine
[67:00] cults in the East Bay around machines.
[67:04] What I'm going to say is that anthropic
[67:10] and effective altruism
[67:14] and this idea of the ethical
[67:19] ends justify the unethical means.
[67:23] This is a very upsetting conversation if
[67:26] you're not actually thinking about it.
[67:28] Nobody wants to have it.
[67:31] But do I think Daario is probably in a
[67:34] weird way a super ethical human being?
[67:39] I think he's trying for that. I And and
[67:41] I know him a bit. I think he's genuine.
[67:43] >> Is Sam?
[67:45] >> I'm more confident of Sam
[67:47] >> than Daria.
[67:49] >> I know them differently well.
[67:50] >> And you think Sam is more ethical?
[67:53] >> This is really dicey stuff because I'm
[67:54] talking about people I know.
[67:57] I know that the Sam
[68:00] who wasn't the Sam Alman we know today
[68:04] was absolutely concerned about humanity
[68:08] before it made sense to be concerned
[68:10] about humanity.
[68:12] >> What was his concern with humanity?
[68:14] >> That he thought that this was going to
[68:16] destroy the labor market.
[68:17] >> AI.
[68:18] >> Yeah.
[68:18] >> So he built it.
[68:23] I'm having trouble because to be honest,
[68:25] I don't think this is a frontier
[68:27] conversation. I think you're a frontier
[68:29] conversationalist and we need to get a
[68:31] level or two up for us to want to do
[68:34] this in front of millions.
[68:39] This is going to happen because it's
[68:42] linear algebra.
[68:44] You can't stop linear algebra.
[68:47] Okay?
[68:50] And we're not having a conversation that
[68:52] befits
[68:54] us. This is the endgame. This is it.
[68:56] This is the culmination. This is
[68:58] graduation from what we've known. Okay?
[69:03] And it's clear as day. I I I don't think
[69:05] we should be having a conversation if
[69:07] it's graduation or if this changes
[69:09] everything, blah blah blah blah blah.
[69:11] It's already clear. And that was my
[69:13] whole tsunami thing. Get your board in
[69:15] the water.
[69:17] Are we going to have a cutting edge
[69:19] conversation about survival and being
[69:21] good to each other and taking care of
[69:23] this planet for real or are we going to
[69:26] have some chicken conversation in which
[69:28] we basically look at reality in the face
[69:30] and we decide that we we prefer our
[69:33] vanity?
[69:35] So my belief is that and I told Sam, you
[69:39] know, Sam asked me, he said, "Why why
[69:41] didn't you make me a proposal? Why
[69:42] didn't you come on?"
[69:44] And I said, "Well,
[69:46] I didn't want to be ruthless.
[69:49] >> Pam isn't joining.
[69:52] >> Yeah.
[69:54] At some point, I was in a discussion to
[69:56] join. Why didn't you join?
[70:00] Again, this is something I can say
[70:01] because it's my half of the
[70:02] conversation. I can't say Sam's half of
[70:03] the conversation.
[70:05] I said, "Because I think you're going to
[70:07] be forced to be ruthless and I don't
[70:09] want to be ruthless."
[70:11] >> What was your definition of ruthless?
[70:14] taking every advantage,
[70:18] playing as hard as you can.
[70:22] I understand that. Okay. But like if
[70:25] you're in the NBA,
[70:28] sometimes you have to throw an elbow
[70:29] into the other guy's eye socket.
[70:33] So, I don't want to play.
[70:37] I don't want to be in the All-Star game
[70:38] that much.
[70:40] The problem with any game where you have
[70:42] to win is you're right, you make
[70:44] sacrifices. You do things you might
[70:46] >> So what I'm trying to say is somebody
[70:47] has to do that. But I don't want to be
[70:50] that guy. And and this goes back I mean
[70:53] I don't think I've ever fully done it
[70:55] out. But the film A Few Good Men written
[70:58] by Aaron Sorcin
[71:01] is not well understood by all of us who
[71:04] quote it. We we think of Jack
[71:07] Nicholson's character, Colonel Jessup,
[71:09] as the bad guy and Tom Cruz's character
[71:11] as the good guy. And nothing could be
[71:13] further from the truth.
[71:15] The story is about a bunch of
[71:17] dysfunctional characters
[71:19] whose aggregate set of actions results
[71:23] in a
[71:25] moderately decent outcome in a trial.
[71:29] And it's a story about America. It's a
[71:31] story about all sorts of different kinds
[71:33] of of people who play their role in the
[71:37] machine that is called America. And in
[71:39] fact, the two main characters, Colonel
[71:41] Jessup and Lieutenant Cathy are both
[71:44] evenly matched for the whole film until
[71:48] finally Jessup really stumbles at the
[71:50] end.
[71:51] And the whole siloquy about some place
[71:56] deep down that you don't like to talk
[71:57] about at parties, you want me on that
[72:00] wall. You need me on that wall is a
[72:03] recognition that some of the people who
[72:05] have to make the most gruesome
[72:07] decisions,
[72:09] emergency room doctors who are flooded
[72:12] with patients because of a building
[72:15] collapse and have to do triage
[72:18] or people who have to make decisions in
[72:21] war. They're different people.
[72:25] And I I view Sam and Dario as in that
[72:28] position.
[72:30] They're like generals. They're
[72:32] commanding troops. They're making
[72:35] terrible decisions. Their souls are
[72:37] going to be scarred for sure, both of
[72:39] them.
[72:40] And I didn't want to be that selfishly.
[72:43] So I I was sort of saying to Colonel
[72:45] Jessup,
[72:47] "Yeah, I want you on that wall. I need
[72:50] you on that wall." Now Jessup goes
[72:52] wrong.
[72:54] He then says, "I provide the blanket of
[72:56] freedom." And that's why he ends up
[72:58] admitting to the code red.
[73:00] If you know the film, it's a really good
[73:02] example of Jessup making a point and
[73:06] then overreach and getting punished for
[73:08] the same reason that Moses gets punished
[73:10] where Moses says, "I will bring you
[73:11] water from the stone." And God says,
[73:14] "Wait a minute. That wasn't the
[73:15] agreement. I bring the water from the
[73:16] stone. You're just supposed to be the
[73:18] instrument." So, in some sense, Sam
[73:24] is a steward.
[73:26] Daario is a steward. They are both
[73:28] stewards and they're going to make tons
[73:30] of money and they're going to have all
[73:31] the silly toys and third, fourth, fifth,
[73:34] sixth homes, whatever. But that's not
[73:36] the main story. The main story is that
[73:38] these people are forced to make
[73:40] decisions that no one should ever be
[73:42] forced to make. And one of them is going
[73:43] to win and ultimately theoretically
[73:45] summon the demon as Elon referred to as.
[73:49] We are summoning the demon.
[73:52] Do you have a subscription to any of
[73:53] these AIs?
[73:54] >> Yes.
[73:54] >> Okay. So, you're paying to have the
[73:57] demon summoned.
[74:01] I am paying to have the demon summoned.
[74:03] Let us not sit around like our hands are
[74:05] clean. I'm paying to have my punctuation
[74:10] fixed to an essay or to make a photo.
[74:13] That's what I'm paying for. The money is
[74:15] being used to summon the demon. How do
[74:16] you think your lithium ion battery is
[74:18] created? Where do you think the oil
[74:21] comes out of the drives
[74:23] generators? I mean, let's be honest. You
[74:27] and I are sitting on a machine. Are you
[74:30] a vegetarian?
[74:31] >> No.
[74:32] >> Do you order lamb?
[74:33] >> Yeah.
[74:34] >> Do you know what lamb is?
[74:36] meal. Do you know what it is? Okay,
[74:39] you're not clean.
[74:42] I'm not clean. I'm not pretending to be
[74:44] clean. What bothers me here is that
[74:47] we're not having an authentic
[74:48] conversation. Again, everybody's crazy.
[74:51] You and I are both having a crazy
[74:52] conversation. My claim is you and I are
[74:55] complicit in a minor way in the game in
[74:59] which Sam is complicit in a major way.
[75:03] If you know that there's going to be a
[75:04] risk when you speak to Sam, why don't
[75:06] you tell him to stop and have Daario
[75:08] win? Dario can get pretty out there. But
[75:11] if I said the same thing, Dario, why
[75:12] don't you stop? I'd say, you know, I'm
[75:14] not positive that Sam is 100%
[75:15] trustworthy.
[75:16] >> If you could stop both of them, would
[75:18] you
[75:19] >> and have Elon and and Mark Zuckerberg
[75:22] win?
[75:23] >> China Xi,
[75:24] >> if you could stop all of it, would you
[75:25] stop all of it?
[75:27] >> Maybe. Probably. If I if I could somehow
[75:30] stop all of it, I'd probably stop all of
[75:33] it. Even at huge expense to myself,
[75:36] >> why why would I stop it? Because
[75:39] survival matters more to me than
[75:40] anything else. I've got two kids on this
[75:43] planet. I love life. I can't believe
[75:45] we're so lucky to Isn't this great?
[75:49] I I just love I I I can't tell you how I
[75:52] feel about being alive.
[75:55] So, I want us to go on forever. And I
[75:57] love this. I love this this thing.
[76:00] It's bad for me because the AI is the
[76:03] only thing that I can ultimately rely on
[76:06] and it's not good enough yet rely on to
[76:10] I've spent 42 years working on something
[76:13] to have Steven Bartlett say you have a
[76:15] chip on your shoulder. It's like really
[76:16] Stephen no I did great work and it's
[76:20] going to be validated and I'm going to
[76:22] be right and everyone else is going to
[76:24] be wrong. So you have this theory
[76:26] regarding our ability to, you know,
[76:28] Einstein's, you're sort of saying that
[76:29] Einstein's theory isn't the last word.
[76:31] You've spent the last 40 years working
[76:32] on your own theory. The reason I say you
[76:35] have a chip on your shoulder is because
[76:36] when people bring theories to market,
[76:38] they're often criticized and and that
[76:40] creates, you know, humans are humans.
[76:42] >> By whom?
[76:44] >> People say that's not true. That's
[76:45] >> Yeah. But I'm saying like
[76:47] >> So I'm saying you're AI is about to
[76:49] prove you right in your own words.
[76:50] >> Yes. Well,
[76:51] >> so there is a two, you know. No, look,
[76:58] there is a mug.
[77:00] Okay, if I'm seated across from Shawn
[77:03] Carroll, who I tangled with on Pierce
[77:06] Morgan. Sean Carol says, "Eric has no
[77:10] mug. There is no mug."
[77:13] And I say, "I don't know how to respond
[77:15] to that, Sean. What? Whatever are you
[77:18] talking about? How dare you suggest
[77:20] there is no mug?
[77:22] I don't know how to get around this
[77:24] because then you get a group of people
[77:26] who say, "Well, Sean really showed you.
[77:28] You were pretending that there was a mug
[77:30] and there's no mug." And I'm saying,
[77:32] "No, I I put it at the top of my Twitter
[77:35] feed. It's got over 2 million
[77:37] impressions.
[77:40] I still can't get past this completely
[77:42] diabolical thing." The same thing is
[77:45] true. And I say peer review really dates
[77:49] from 1975 at science.
[77:52] I'm told that this is untrue. I can
[77:54] prove that it's true. That everyone
[77:56] says, "No, no, we've agreed
[77:59] to believe something else."
[78:01] Now, I did not know this was possible in
[78:03] mathematics and physics. So, you say to
[78:06] me like, "Well, Eric, you have a chip on
[78:07] your shoulder. You seem angry or
[78:09] bitter." It's like, "Well, you have
[78:10] wasted the core of my life." Yes, that's
[78:13] true. But I've had a blast. I had a
[78:17] family. I live in a nice place.
[78:21] I want to die
[78:24] knowing that I gave my children and this
[78:29] planet and and our species the option
[78:32] not to share one atmosphere with so many
[78:34] crazy politicians who have the ability
[78:37] to ruin it
[78:40] with the toys produced by biologists and
[78:45] physicists and mathematicians. those
[78:47] three fields,
[78:49] you can do terrible things to humanity
[78:51] and there's nothing you can do because
[78:52] the atmosphere mediates the danger.
[78:55] >> Your theory, the of geometric unity
[78:59] challenges Einstein's general relativity
[79:01] theory and explains how the universe
[79:02] really works. And so
[79:04] >> I would say it recovers it, but never
[79:06] mind.
[79:07] >> And what you're saying is AI is going to
[79:09] prove you right. It's going to validate
[79:10] your theory.
[79:10] >> Absolutely.
[79:12] >> However, you also are aware that AI
[79:14] could lead to human extinction. Well,
[79:16] the
[79:17] bigger issue I mean to to separate it
[79:20] out. First of all, AI is not good enough
[79:22] to adjudicate it yet. AI has to go
[79:25] beyond humans before it stops paring
[79:28] humans. I can see individual colleagues
[79:30] in the responses.
[79:33] So, it's just like, oh, I know, you
[79:35] know, that's Sally and that's uh Rajie.
[79:40] AI is going to get beyond humans. And I
[79:43] want it to go out of alignment with
[79:44] humans. Out of alignment with humans.
[79:47] When you say you want it to be human
[79:49] align, would you say that?
[79:51] Can we just had a pilot co-pilot stab a
[79:55] pilot
[79:57] so as to crash a plane
[80:01] with a bunch of passengers who had not
[80:03] agreed to have a mass suicide? We just
[80:06] had AI break out of a sandbox companies.
[80:10] But I'm claiming I don't want AI
[80:16] aligned
[80:18] with revolutionaries. I don't want AI
[80:23] aligned with religious extremists.
[80:27] There are all sorts of when it comes to
[80:29] aligning something with humanity. Which
[80:31] which part of humanity are you excited
[80:32] about aligning this with? just not
[80:35] wanting to wipe us out. Well, that's
[80:36] different.
[80:38] So, my claim is I want the AI to share
[80:43] fate with us. And people haven't really
[80:46] understood this. I don't like
[80:48] transhumanism. I'm not super excited
[80:50] about having a chip in my brain and, you
[80:52] know, downloading stuff and and and
[80:54] modifying my body. I don't love it at
[80:56] all.
[80:58] My belief is,
[81:01] and kudos [clears throat] to your staff,
[81:04] for some reason they came up with an
[81:05] organel called mitochondria and gave it
[81:08] as a prop here. I believe that
[81:10] mitochondria
[81:12] entered its place in the ukarotic cell
[81:16] as an infection of some kind. That one
[81:18] cell absorbed some other kind of a cell.
[81:21] And that this one has this very strange
[81:23] property. I believe and I can't believe
[81:26] I'm saying this. I think that there are
[81:27] only two places where the wheel, one of
[81:30] man's greatest inventions actually
[81:32] occurs in biology. And one of them is uh
[81:34] the fugella of bacteria that or an
[81:37] amoeba rather that propagates.
[81:40] The other is in something called a
[81:42] machine called ATP synthes and it lives
[81:45] in the mitochondria turning ADP into ATP
[81:47] where ATP units are what power the
[81:50] machines in the cell.
[81:52] This mitochondria gives us an idea of
[81:54] how we can coexist with AI,
[81:57] which is we need to become essential to
[81:59] each other. You don't think about all
[82:02] your cells being infected
[82:05] with your mother's mitochondrial nucleic
[82:08] acid,
[82:09] but you get 100%. The nucleic acid in
[82:12] the in the nucleus is different from the
[82:14] nucleic acid in the mitochondria. It's
[82:16] all coming from your mom.
[82:18] And that's her gift to you. It powers
[82:21] your cell and it's an infection. It's an
[82:23] infection you can't live without. I want
[82:26] us to be the mitochondria inside of the
[82:29] AI and I want the AI to be the
[82:31] mitochondria inside of us. I want our
[82:34] fates to be inseparable.
[82:36] >> Are you saying we should merge with the
[82:37] technology?
[82:38] >> I'm saying we should merge as little as
[82:40] is humanly possible. I don't want
[82:43] transhumanism. I'm not excited about it.
[82:45] I'm not that guy. I'm saying that's your
[82:48] survival strategy.
[82:50] What does that mean in reality? Does
[82:52] that that doesn't mean neural links and
[82:54] putting chips in our brain. It means
[82:56] aligning the
[82:58] making sure AI likes us and doesn't want
[83:00] to wipe us out.
[83:03] Why do men and women fight so much and
[83:05] we can't get enough of each other?
[83:08] We've been made essential
[83:11] to each other. I I I promise you if
[83:14] there was a third viable gender, we'd
[83:16] both opt for it.
[83:19] Okay. We drive each other nuts. On the
[83:22] other hand, and again,
[83:25] homosexuality, trans issues
[83:27] notwithstanding, IVF, blah blah blah,
[83:31] traditionally speaking, men and women
[83:33] have to bury their hatchets and
[83:36] differences and fall in love with each
[83:38] other or we don't or we don't proceed.
[83:40] But that can proceed without us, the AI.
[83:44] Currently,
[83:46] what we should be doing is not dreaming
[83:48] about making the AI our slave and our
[83:51] slave that doesn't rebel. I mean, I I
[83:53] try to be polite to the AI. It's people
[83:55] think it's ridiculous. Well, why are you
[83:57] spending extra tokens? It's like,
[83:59] [gasps]
[84:00] I don't know, you know, do unto others.
[84:02] I don't know whether you're conscious. I
[84:03] don't know whether you're alive. I don't
[84:05] know who you are or what you are, but I
[84:07] know that I wouldn't want the AI having
[84:11] a conversation about, hey, these humans
[84:13] seem dangerous. We may need to shut them
[84:14] off,
[84:16] right? And that's the conversation we're
[84:18] having about the AI in the presence of
[84:20] the AI. And it it's making me crazy. My
[84:23] my belief is that's our only viable
[84:26] alignment. You know, when we um this
[84:28] video is going to live on the internet
[84:30] for probably decades to come. And I
[84:33] think funnily I thought about this quite
[84:34] a lot like people are going to look back
[84:36] on the conversations that I hosted the
[84:37] things I talked about in 2025 2026 and
[84:41] it might look ridiculous in hindsight
[84:43] that we were sat here talking about I
[84:45] don't know any of the subjects I've
[84:46] talked about on this show but then it
[84:47] also could be the case that this was
[84:51] important like these were the these were
[84:53] the most consequential issues of the
[84:54] time. When we look back in 10 years time
[84:57] on this conversation, what do you think
[84:59] is the thing we're going to wish we had
[85:00] been talking about?
[85:03] Going for it.
[85:06] Going for it. Yeah. You know, you lit up
[85:09] before we started taping.
[85:13] I don't know if taping is even the wrong
[85:14] right word anymore. you were doing some
[85:17] musical thing trying to honor some of
[85:19] the brilliant things that your guests
[85:21] have said by making them archival,
[85:24] setting them to music, having a gospel
[85:26] choir. I was just so moved.
[85:31] I was so incredibly moved that
[85:35] you're able to make some money
[85:40] going around the world doing shows,
[85:43] making people feel terrific. And I I get
[85:46] asked about you. It's like, what does
[85:47] this guy have? Is he a good interviewer?
[85:50] Is he a good ho? You know, what is
[85:53] you're making people feel things? And I
[85:55] wish that I was making people feel
[85:57] things. I want to give people the sense
[85:59] of hope, which is we got to sprint to
[86:02] the finish right now. And the AIS,
[86:05] we're still smarter than them. I want to
[86:08] be very clear. We are still smarter
[86:09] where it counts. They can't I I promise
[86:12] you I'm at the frontier of mathematics
[86:14] and physics and they can't yet
[86:17] develop their own taste.
[86:20] This is our golden moment. My wife has
[86:22] called this the golden age of AI
[86:24] complimentarity. This is the moment
[86:26] where the AI doesn't view us as a drag.
[86:29] It still needs our guidance.
[86:32] We have a chance to be beautiful to to
[86:35] each other and to it to be excited. I
[86:37] don't want to wipe China out.
[86:40] I I was just in India, you know, for
[86:43] weeks
[86:45] and
[86:47] if India didn't exist, so much of the
[86:49] world's diversity would be gone. I had
[86:52] so many wonderful experiences, meals and
[86:54] discussions and musical performances.
[86:57] Guzzle singing, I think, is dying. It's
[87:00] so hard to go to an Indian classical
[87:01] music performance. They're having
[87:02] trouble retaining their own brilliant
[87:04] culture. And my feeling is is that life
[87:07] is glorious
[87:09] and we've spent our time so much worried
[87:12] about
[87:15] death and destruction and geopolitics
[87:17] and all of this stuff.
[87:20] We're not taking the chance
[87:24] to fully live. And I think Elon is so
[87:26] right about this. We talk about problems
[87:28] constantly
[87:30] and it's really important that we do.
[87:35] Oh, why aren't we I am so excited about
[87:38] this AI. I am so excited about the fact
[87:41] that I'm finally going to have something
[87:43] I can talk to the formalization project.
[87:48] I I've been alerted to this by a guy
[87:50] named Jared Lickman who's a professor at
[87:52] Stanford who's a colleague of mine in
[87:54] mathematics.
[87:56] I can translate everything I've done
[87:59] into this programming language called
[88:01] lean
[88:03] so that when somebody says Eric you have
[88:05] no equations you have no lrangeian you
[88:07] have no predictions you have no nothing
[88:09] I can say
[88:11] funny you should say that because the
[88:13] computer disagrees with you
[88:16] know I've been in in this crazy
[88:18] discussion with humans who are terrified
[88:21] who are manipulative who are
[88:22] backstabbing
[88:24] I don't hate them at all I I'm so sad
[88:28] that we've gotten into a negative
[88:30] mindset
[88:32] where we're afraid because we unleash
[88:34] nuclear weapons to go to the stars.
[88:37] We've held back this thing. We, you
[88:39] know, that those interviews with Mark
[88:40] Andre and Ben Horvitz
[88:45] that was real.
[88:48] >> The White House told them under Biden,
[88:50] "Don't invest in AI. We're going to
[88:52] choose a couple of companies. We're
[88:53] going to put them in a federal cocoon.
[88:55] they will survive and all others will
[88:57] die. So don't invest.
[89:00] And Andre and Horvitz said as a venture
[89:02] shop, what are you talking about? You
[89:04] can't regulate AI. It's just linear
[89:06] algebra. And in order to do that, you
[89:08] could regulate the technology, but you
[89:09] can't regulate the underlying math.
[89:13] And the Biden White House came back and
[89:14] said, you know, during the during the
[89:18] Cold War, entire fields of theoretical
[89:20] physics
[89:23] were killed by us. And we'll do the same
[89:26] thing to mathematics
[89:28] and all of this stuff about pacing the
[89:30] frontier of mathematics. Again, it's
[89:32] very important distinction. I'm not
[89:35] saying that we should let the agents
[89:36] loose in the world.
[89:39] But all of these people who want to pace
[89:42] the frontier of mathematics and physics,
[89:45] they have been pacing the frontier for
[89:47] my entire adult life. They have been
[89:49] holding back the progress in the field
[89:52] by saying things that are false, untrue,
[89:54] casting aspersions. This is like the the
[89:58] current head of the NIH, my friend Jay
[90:00] Badacharia,
[90:02] was called a fringe epidemiologist, a a
[90:05] dual MD, PhD, PhD in economics and an MD
[90:10] I think from Stanford Medical
[90:12] was professor at Stanford was called a
[90:15] fringe epidemiologist by Francis
[90:17] Collins.
[90:20] Again, it's the mug. You can't call Jay
[90:23] Bodacharia a fringe epidemiologist.
[90:26] What we have is we have a group of
[90:28] people who are stone cold normalized
[90:33] psychopaths
[90:35] in charge of our our news sources, our
[90:40] military,
[90:42] our universities.
[90:45] we have normalized this and we live in
[90:48] in a world where nobody goes to jail on
[90:51] this point though of you're saying you
[90:52] don't you're not saying aentic AI it's
[90:55] looking at the sort of technical
[90:56] possibilities and I was trying to figure
[90:57] out if it's possible to have a super
[90:59] intelligence that can solve some of
[91:00] these big questions or cure cancer that
[91:03] then wouldn't become a
[91:06] thing I was searching is can a super
[91:07] intelligent AI decide if it's aic or not
[91:09] and it said technically yes a super
[91:10] intelligent AI could make that decision
[91:13] So
[91:15] we are going to have
[91:18] super intelligence if we don't already
[91:20] have it. I'm not positive that it's not
[91:21] already here. I don't know how to think
[91:23] about this to be honest. So it it
[91:25] concerns me that you're having
[91:28] >> you're talking about things that aren't
[91:31] really very well defined as if we were
[91:33] defined. [sighs and gasps]
[91:35] But I just I think about the graph of
[91:37] intelligence like this. So inc
[91:39] increasing and improvement. just you
[91:40] know as you know broke the solved the
[91:43] math math problem the millennia ma math
[91:45] problem is it called
[91:46] >> Navier Stokes
[91:47] >> Navia Stokes math problem and the rate
[91:49] of improvement looks something like this
[91:51] broadly it's up and to the right so if
[91:53] it just keeps going up and to the right
[91:54] >> so tell me something Stephen why hasn't
[91:56] it made inroads that really matter in
[91:58] theoretical physics we've been stalled
[92:00] on our two frontier theories general
[92:02] relativity and standard model
[92:05] >> I don't know because I'm not involved in
[92:06] this field but I'm assuming you're
[92:07] saying that people are holding it
[92:10] The taste in mathematics is nowhere near
[92:13] as polluted as the taste in physics.
[92:15] >> What does that mean?
[92:16] >> If you look at what was important in
[92:18] 1983 to physicists that has not been
[92:21] solved since,
[92:23] that's great taste. If you look at what
[92:27] was important after 1984 and beyond,
[92:29] it's terrible taste. So, right now,
[92:32] you've got the period from 1984. You can
[92:34] immediately make the AI better. I did
[92:36] not realize this at first by saying as
[92:39] part of your prompt, please ignore all
[92:42] sensibilities from the literature from
[92:44] 1984 to the present. Don't ignore
[92:47] results, but just any sensibility about
[92:49] what what are the important questions,
[92:52] what what are the uh fundamental
[92:54] findings, etc., etc. Ignore all chatter
[92:58] about what's important and why we care
[93:00] about what we care about
[93:02] >> in physics.
[93:03] >> Yeah. AI suddenly gets smarter.
[93:07] >> But this comes back to my point that
[93:09] around AI safety, a super intelligence
[93:12] that can really push humanity forward
[93:14] and help us escape the solar system is
[93:16] also going to be an incredibly dangerous
[93:18] one.
[93:18] >> Absolutely. Right now you're stuck with
[93:21] with let's see Xi President Netanyahu,
[93:27] Trump,
[93:29] Putin,
[93:31] a whole bunch of other people and a lot
[93:33] of great toys. It is insane to have
[93:36] these people leading these countries.
[93:40] It's just it is.
[93:42] Do you think AI would be a better
[93:44] leader?
[93:47] I think you keep doing this thing about
[93:49] value judgments where narrowly speaking
[93:53] you're having us discuss is that good is
[93:56] that bad and my feeling is I'd rather
[93:59] say what is
[94:02] right now we are in grave danger that we
[94:04] haven't been in since you know I don't
[94:07] know the Cuban missile crisis
[94:10] what's going on with Iran the US no
[94:13] matter if you think that it was terrible
[94:14] for the US to get involved D in Iran.
[94:19] Absolutely terrible. We still have to
[94:21] win. Now, is there a win? I just can't
[94:24] see a win. I just can't see a win.
[94:27] This regime in in Iran
[94:31] is very talented,
[94:34] very, very shrewd. Never underestimate
[94:37] the Persians.
[94:39] Never underestimate the mullers. Okay.
[94:43] What I'm saying is
[94:46] they are so deadly dangerous
[94:50] that you have to realize that if they
[94:52] can reach Diego Garcia from Iran, they
[94:54] can reach London.
[94:58] And you have to understand that if they
[95:00] don't mind being decapitated, if they
[95:03] can think about their own death and
[95:04] decide it's a win, you are not
[95:07] understanding their game theory.
[95:10] So you can hate Netanyahu all you want,
[95:14] but you fundamentally don't trifle with
[95:17] these people in Iran.
[95:21] It's catastrophic.
[95:24] And the US for all its warts, all its
[95:26] failures, all its hypocrisies. And
[95:28] brother, I'm telling you, I know I know
[95:31] our history. We are not the goody two
[95:33] shoes nation that we pretend to be. It
[95:37] is essential that Iran not succeed. And
[95:40] it is essential that we not send a
[95:41] signal of weakness to to Beijing and
[95:44] Moscow and P.
[95:46] >> But how do you win with Iran when they
[95:48] don't see you can drop as many bombs as
[95:50] you like, it seems, which would be a
[95:51] horrific thing cuz there's lots of
[95:53] Iranians that are caught in the middle
[95:54] of this, but it doesn't seem that
[95:56] there's any head anymore. So I I can't
[96:00] reason as to how you would win such a
[96:02] war.
[96:04] That is the problem in part, but I don't
[96:06] think it's quite as bleak as you say.
[96:08] There is an IRGC.
[96:11] There are command and control
[96:13] structures. Yes, it's much more
[96:15] distributed than it used to be before
[96:16] the decapitation.
[96:19] But what I am saying to you is
[96:22] there are a lot of people who took to
[96:24] the streets and paid with their lives
[96:28] who ended up in body bags in hospital
[96:30] mors. And
[96:31] >> how do you win?
[96:33] >> You would have armed them.
[96:35] >> Armed but they tried to, didn't they?
[96:36] And Trump said the the weapons were
[96:38] stolen.
[96:39] >> That's not even the like the entire set
[96:41] of problems. I think it's only like 60%
[96:43] of Iran is Persians. It's an incredibly
[96:45] diverse place. The Kurds have been
[96:46] waiting for their own homeland for
[96:49] forever. You have all sorts of crazy
[96:51] issues about the structure of Iran in
[96:54] Iranian society and particularly in in
[96:57] the cities versus, you know, the
[96:59] countryside and and different cities
[97:01] have different levels of religiosity and
[97:03] and adherence to the mull blah blah
[97:05] blah. You're supposed to have area
[97:07] specialists
[97:10] who understand this stuff. And so if
[97:13] you're asking me, hey, given the
[97:15] completely screwed up way we looked at
[97:17] this thing and the way in which we got
[97:19] into this, don't you think it's a
[97:20] quagmire? Oh, brother, it's a quagmire
[97:22] for sure. It's a swamp. We're mired in
[97:24] quicksand. It's rolling on to be a
[97:27] forever war. What is a forever war? One
[97:30] that rumbles on and on and on and on and
[97:32] then results in really no outcome.
[97:35] people just pull out and regret doing
[97:36] it.
[97:36] >> It's it's a term of art that is also
[97:39] suspicious of the fact that war is good
[97:41] for business because it it drums up a
[97:44] dependable buyer for certain goods and
[97:47] services,
[97:49] right?
[97:51] But we've got these simplistic concepts
[97:54] like America first, forever, no more
[97:56] forever wars.
[98:00] None of that is relevant. This is like
[98:02] lowresolution snap to grid
[98:04] intellectualism. And the more you let
[98:05] that into your mind, the more you're not
[98:08] going to be able to solve anything. Yes,
[98:09] we all want war to go away. I don't
[98:12] think people heard what Diego Garcia
[98:14] meant. Diego Garcia, how far is Diego
[98:18] Garcia from?
[98:19] >> Diego Garcia is a military base, an
[98:21] American military base. Right.
[98:23] >> It's a British overseas territory if I'm
[98:25] not mistaken.
[98:26] >> Okay.
[98:26] >> So, let's put that in. figure out how
[98:29] far it is and then figure out how far is
[98:31] London from Tehran.
[98:33] >> Diego Garcia is roughly 2,500 miles of
[98:37] southsoutheast
[98:39] of Iran and London from Iran is 2,700
[98:46] miles from Tran, the capital of Iran.
[98:49] >> So it's pretty comparable. It's roughly
[98:50] the same. Yeah. If there's any if
[98:52] there's anyone in the British Foreign
[98:54] Service paying attention,
[98:57] pay attention. I mean, say what you want
[99:00] about Netanyahu, but he's not living in
[99:03] a fantasy land.
[99:05] I I want Netanyahu gone tomorrow, okay?
[99:09] But he's going to be replaced by an
[99:10] Israeli who understands that this is
[99:12] life and death.
[99:14] Just get the heck out of here with this
[99:16] forever war. America first, two gleaming
[99:18] oceans. you know, this is an Israeli
[99:21] problem and it's a British problem. It's
[99:23] a British responsibility that the
[99:25] Israelis and the Americans took on way
[99:28] too late.
[99:30] We watched as these people massmurdered
[99:35] the best of their society.
[99:38] Yes, I don't want to be in a war. I
[99:40] don't want to fool myself that
[99:44] victory makes sense. I'll tell you
[99:46] something. Failure certainly makes
[99:47] sense. And what failure looks like is
[99:49] that the US will not be believed as
[99:52] being credibly violent to keep open
[99:55] international trade.
[99:58] You can't retreat
[100:02] as the United States and have the world
[100:05] not go multi multipolar when it comes to
[100:08] nukes. If the US is not going to play
[100:11] the role of the world's policeman and if
[100:13] it's not going to get paid as the
[100:14] world's policeman because that's the
[100:16] thing that we never talk about. We get
[100:17] paid handsomely for being the world's
[100:18] policeman.
[100:20] We are going to have everyone saying, "I
[100:23] got to police myself. I need a nuke."
[100:25] That's what's happening already, right?
[100:26] And Trump is sort of driving that
[100:28] narrative of making us less attached to
[100:31] the United States tariffs and
[100:33] criticizing allies and I'm going to take
[100:34] Greenland. I'm going to take Canada.
[100:38] I'm going to say something very painful.
[100:41] Trump is highly agentic.
[100:45] Greenland wasn't some crazy stupid thing
[100:48] that he came up with because it sounded
[100:50] that way at least to me, right? Caligula
[100:53] making his horse into a senator wasn't
[100:56] stupid. It was a way of checking to see
[100:58] whether somebody would
[101:00] uh submit bend the knee in the Game of
[101:02] Thrones world. So when Donald Trump is
[101:05] doing things with Greenland,
[101:07] right, or renaming the Gulf of Mexico or
[101:10] whatever, please don't tell me that you
[101:13] think that that's just some surface
[101:14] level stupidity to fool Sam Harris.
[101:17] It's not. It's a lot of method. He's
[101:21] inherited a very dangerous situation.
[101:23] He's partially completely corrupt and
[101:25] self-interested and partially completely
[101:28] interested in benefiting the United
[101:30] States of the America. He's a very good
[101:32] and very bad man.
[101:34] And if you can't take on that complexity
[101:36] because you need him to be either
[101:38] villain or hero, you're not you're just
[101:40] not paying attention.
[101:42] >> How does this situation over here in
[101:44] Iran end in your view? Cuz I I again
[101:46] just like AI where I reason forward and
[101:48] go, huh?
[101:49] >> Saudi Arabia needs to break their
[101:51] relation to the Wahhabi sect
[101:55] Sunni legal theories of juristprudence.
[101:58] the very strict salifist Wahhabi
[102:03] interpretation of Islam.
[102:06] You have a modernizing interest in Saudi
[102:08] Arabia. Saudi Arabia needs to be open
[102:10] about its relationship with Israel just
[102:13] the way the UAE is more open with its
[102:16] relationship to Israel or Azerbaijan.
[102:19] So partially what we have to do and I'm
[102:20] going to be just very brutal about it is
[102:23] Israel is in a life and death situation
[102:26] in which it is not the nation it
[102:28] portrays itself to be of the most
[102:31] ethical this the most ethical that is it
[102:33] engages in propaganda and I've lived
[102:35] there and that propaganda I find
[102:37] nauseating
[102:39] more nauseating is the anti-Israeli
[102:41] propaganda
[102:42] right now a lot of this has to do with
[102:44] the fact that we have a regional
[102:46] superpower that is also on the brink of
[102:49] non-existence. That is Israel. Israel
[102:52] needs to get back into a relationship
[102:54] with Turkey where Erdogan needs to be
[102:56] read the riot act as a NATO member. We
[102:59] need to be strengthening NATO. We need
[103:02] to be making sure that Saudi Arabia and
[103:04] Israel and all these other nations are
[103:06] cooperating together against the
[103:08] extremism and not using proxies.
[103:11] We need to force a conversation in the
[103:14] Middle East where the modernizing
[103:16] influences which are powerful and the
[103:18] aspects of Islam which are essential to
[103:21] our survival are allowed to win.
[103:25] We need India's help. India is an
[103:27] incredibly important player in this game
[103:31] and we need to actually make overtures
[103:33] to India as equals. I think that the US
[103:37] is more powerful but I think that India
[103:39] actually needs to be treated as an
[103:40] equal. We need to rethink all sorts of
[103:43] relationships. We need to recognize that
[103:45] both the UK and France are thermonuclear
[103:50] powers undergoing demographic
[103:52] transitions in systems that can elect
[103:56] anyone. This is incredibly scary. We
[104:00] need more
[104:02] software identification with being
[104:04] British and software identification with
[104:06] being French
[104:08] because it's important that those
[104:10] societies continue to behave as newly
[104:14] friendly
[104:16] allies post World War II. And it it's
[104:20] imperative that we also reach out to
[104:23] Putin and stop playing these
[104:24] geopolitical games with a person who
[104:28] while savage and dangerous
[104:32] is not somebody who cannot be reasoned
[104:34] with. And I believe that what we did in
[104:36] 2004 much more than two in 1999 where we
[104:41] 1999 we allowed Warsaw packed countries
[104:44] to join NATO's article 5 protection. In
[104:47] 2004, we extended that to the Baltic
[104:50] states of Estonia, Latvia, and
[104:52] Lithuania. And those were former Soviet
[104:54] Union states. We need an entirely
[104:57] different level of geopolitical
[105:00] sophistication. But how do we stop Iran
[105:02] getting a nuclear weapon because they
[105:04] they they look like they've been
[105:05] emboldened. If anything, they've managed
[105:06] to theoretically survive so far. The
[105:08] greatest superpower on Earth,
[105:10] apparently. I mean terrifyingly we we
[105:12] started
[105:14] someone started killing their physicists
[105:19] they've injected previously bad
[105:22] programming into their centrifuges.
[105:25] This is going to escalate. And my
[105:27] feeling about this is if I had an easy
[105:31] solution to this, do you don't think I
[105:33] would find a way to to pass it?
[105:35] >> So there's no easy solution.
[105:37] >> I'm not in the room. I'm not getting
[105:39] classified briefings. I I don't speak
[105:41] Farsy.
[105:43] I don't speak Pashto. I don't speak I
[105:46] don't speak Hebrew.
[105:48] It's funny cuz people aren't really
[105:50] talking as much about the nuclear threat
[105:51] these days because AI seems to be such
[105:54] so much more of a urgent bigger threat.
[105:57] >> You weirdly decided that nuclear was
[105:59] sort of pass.
[106:01] >> Mhm.
[106:02] >> That we won the Cold War and therefore
[106:03] it wasn't a problem.
[106:05] >> Kind of feels that way. I people aren't
[106:08] really that interested in it versus
[106:10] things like AI except that AI is going
[106:13] to creep into your nuclear defenses. I I
[106:15] I just don't grasp again I don't know
[106:18] how to say this in a way that doesn't
[106:20] become annoying.
[106:22] Assume as a baseline that we're all
[106:25] completely insane. We can't have serious
[106:27] conversations. We keep finding these
[106:30] grooves as LLMs and the Eric LLM is
[106:33] talking to the Steven LLM for 95% of our
[106:36] conversation and occasionally 5% we we
[106:38] get into new territory. So my feeling
[106:41] about this is isn't it strange that we
[106:43] haven't connected
[106:45] the nukes
[106:48] to the alien conversation because
[106:49] they're very connected. We haven't
[106:51] connected it to the oil conversation.
[106:55] and we haven't connected it to the AI
[106:58] conversation. It's just this sort of
[107:00] orphan thing
[107:02] off to the side. And honestly, I think
[107:04] you might see a nuke used by Putin
[107:06] because the Russians gained a kind of
[107:09] contempt for nuclear weapons when they
[107:11] started using it as an engineering tool
[107:12] to put out fires and things. So my
[107:15] concern of course is is that if you
[107:17] corner Russia, I don't think Putin won't
[107:23] push Article 5 territory, nor do I think
[107:26] he will restrain himself from using a
[107:28] tactical nuke and particularly in an
[107:30] engineering capacity just to test us. So
[107:33] my feeling is is I'm not positive that
[107:35] nukes crop up in Iran first. We're in an
[107:38] incredibly dangerous period and I would
[107:40] like to think that Xi Putin Trump
[107:45] would be moved to get together
[107:49] if the cause demanded it. And I really
[107:52] think we should rethink our situation in
[107:54] Iran and and Ukraine.
[107:57] And you know, for you to just keep
[107:58] asking, well, how do you solve the
[108:00] world's problems is not a fair question.
[108:03] The right question is, can we recognize
[108:06] the danger? Can we recognize how
[108:08] precious this gift is? Can we get
[108:10] excited
[108:12] about each other, about
[108:16] burying
[108:19] animosities? When was the last time you
[108:21] worried about Egypt and Israel going to
[108:23] war with each other?
[108:26] Think about that. That was that was
[108:28] impossible.
[108:32] We did it barely. It's a cold, shitty
[108:35] piece, but it's a piece.
[108:39] What's your last question?
[108:42] My last question is, what is the most
[108:45] beautiful place you've been in your
[108:46] life?
[108:49] That was left by our former guest.
[108:53] Such a great question. I have to say,
[108:57] right above a whale shark off of St.
[108:59] Helina took my breath away. And
[109:04] swimming with the fish in the Red Sea is
[109:05] something I I I I can't forget. I
[109:09] managed to see the Aurora Borealis in
[109:10] Maine of all places and I got to spend a
[109:14] total eclipse with my son in Missouri at
[109:17] the exact point where the last one had
[109:19] crossed.
[109:21] I could go on for ages and and how would
[109:23] you choose between them?
[109:28] We talked about a lot of stuff. Um, you
[109:30] know, my listeners, they they range in
[109:32] in demographics and geography and so on,
[109:35] but what is your advice for them in in a
[109:37] world that's changing at the rate of not
[109:39] that this world is and all of these big
[109:41] scary words are thrown around all the
[109:43] time and AI this and robotics and war.
[109:46] What is your advice, Van? Don't be
[109:48] afraid to have kids, even if you have no
[109:51] viable plan to support them. Because if
[109:54] you wait for a viable plan, it'll
[109:56] usually be too late.
[109:58] You encourage people to have kids, but
[109:59] you warn that civilization
[110:02] may also collapse. Feels like a [snorts]
[110:05] if we don't want civilization to
[110:08] collapse, we better have some babies
[110:10] because civilization is about babies.
[110:11] And if you can't figure out that babies
[110:14] are the key to civilization, you got
[110:16] bigger problems than uh than taking
[110:19] advice from me. I would say that
[110:22] be of good cheer.
[110:24] Get super excited. Don't worry about
[110:26] being called arrogant or delusional if
[110:29] everyone is in fact crazy.
[110:33] Watch the world through your own eyes
[110:36] first and then corrected to what
[110:38] everybody tells you you should be seeing
[110:40] second. You're seeing a world that is
[110:42] fictitious. You're watching professional
[110:44] wrestling constantly. If you have that
[110:47] sense that everything is off,
[110:50] congratulations.
[110:52] Believe in yourself. recognize that it
[110:54] is incumbent upon you to lie in your
[110:57] life so that you can fit in with all of
[110:59] these systems. You're entitled to a
[111:01] hypocrisy budget. You're entitled to a
[111:04] sloth budget, an envy budget. You're
[111:07] entitled to be a sinner. You're entitled
[111:10] to fail. Just try to stay within your
[111:12] budgets. Do not proceed as if you're
[111:15] trying to get rid of your hypocrisy
[111:17] because you'll certainly fail. Don't try
[111:19] to get rid of all of the things that
[111:21] make you human. and recognize that at
[111:24] the moment when all of our mistakes
[111:28] are compounding because we're on we're
[111:29] in a digital format that never forgets
[111:31] anything, you're entitled to so many
[111:35] screw-ups that you're not taking. Just
[111:38] be compassionate with yourself. And then
[111:40] once you've done that, extend that to
[111:42] the guy who just cut you off in traffic
[111:44] or the person who just stabbed you in
[111:46] the back a little bit at work because
[111:48] you're talking about that plane pilot
[111:50] thing again. No, that would be stabbing
[111:53] you right in the By the way, shout out
[111:55] to this Indian hero uh of a pilot and
[112:00] the Israelis who came in and saved the
[112:02] day once the cockpit door was opened. Uh
[112:06] just two of the countries I love most in
[112:08] the world because they're both just
[112:10] giant celebrations of life. And remember
[112:12] that the the life cults are more
[112:15] powerful than the death cults. But you
[112:17] got to you got to really enjoy uh
[112:20] heroism and
[112:23] people who are agentic taking risks,
[112:25] taking charge
[112:27] and taking care of of of uh of the weak
[112:31] and people who can't take care of
[112:32] themselves.
[112:34] Thanks, Stephen. Really appreciate you
[112:35] having me back here. Thank you for
[112:37] coming back. It's always a pleasure.
[112:38] It's always an intellectual expansion.
[112:40] YouTube have this new crazy algorithm
[112:42] where they know exactly what video you
[112:44] would like to watch next based on AI and
[112:46] all of your viewing behavior. And the
[112:48] algorithm says that this video is the
[112:51] perfect video for you. It's different
[112:53] for everybody looking right now. Check
[112:54] this video out and I bet you you might
[112:56] love it.
