Cisco's Patel Rejects AI Job Cuts, Warns of Network Bottlenecks
Ai4
Cisco President Jeetu Patel warns that eliminating entry-level roles threatens organizational resilience, while autonomous AI agents will soon overtax existing energy and network infrastructure.
## Background
Published on YouTube by Ai4 on August 24, 2025, this 18-minute video features a fireside chat between Cisco CPO Jeetu Patel and CNN Senior Reporter Matt Egan. According to the page description, the discussion explores the transformative power of agentic AI, organizational adoption risks, and Cisco's approach to securing AI ecosystems at machine scale.
## TL;DR
Cisco's Jeetu Patel and CNN's Matt Egan explore the transition from interactive chatbots to autonomous agentic AI operating at a scale of 10 to 100 agents per person. Patel dismisses claims of total human obsolescence and argues against eliminating entry-level jobs, emphasizing the importance of unconditioned perspectives and meta-prompting techniques. The conversation also addresses the severe power and network infrastructure demands of computer-use agents, alongside the necessity of public-private reskilling and multi-layered security to combat AI-driven cyber threats.
## Atmosphere & style
The session carries an engaging, forward-looking, and intellectual atmosphere, driven by Egan's incisive journalistic questioning and Patel's pragmatic tech-executive perspective. The dialogue balances high-level optimism regarding AI abundance with grounded concerns about cybersecurity, energy consumption, and workforce disruption. Interactive audience polls and personal anecdotes—such as Patel's daughter interacting with a physical newspaper—add a conversational and relatable warmth to the strategic discussion.
## Key takeaways
- [00:04](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=4s) AI introduces a fundamental shift from human history's traditional resource scarcity to an era of potential abundance.
- [00:04](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=4s) AI hype is exaggerated regarding complete human job displacement within five years, but understated regarding AI's ability to generate completely novel, non-human insights.
- [02:59](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=179s) AI is evolving from interactive phase-one chatbots into phase-two autonomous agents capable of performing multi-step tasks over extended timeframes.
- [02:59](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=179s) Scaling agentic AI faces critical bottlenecks in power and compute infrastructure alongside a growing human trust deficit.
- [06:35](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=395s) As intelligence becomes commoditized, competitive advantage shifts toward question formulation and using meta-prompting to optimize inputs.
- [09:36](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=576s) Eliminating entry-level white-collar jobs is counterproductive because early-career workers bring unconditioned perspectives required to unlearn outdated assumptions.
- [15:02](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=902s) Managing workforce displacement requires public-private partnerships for reskilling, as non-tech workers cannot instantly adapt to new AI roles.
- [17:06](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=1026s) Securing AI requires dedicated protections against hallucinations, prompt injections, fine-grained agent identity controls, and machine-scale cyberattacks.
## Detailed walkthrough
### [00:04](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=4s) Rethinking Core Assumptions: Moving from Scarcity to Abundance
The conversation opens with CNN Senior Reporter Matt Egan asking Cisco President and Chief Product Officer Jeetu Patel about his view that artificial intelligence is bigger than anything seen in his lifetime. Asked what fundamental societal assumptions must be reconsidered, Patel explains that human history has always been defined by scarcity—managing limited resources, which in turn drove invention. AI, however, introduces the possibility of true abundance, opening avenues to solve previously intractable challenges. While reaching full abundance remains years away, Patel highlights a critical dichotomy: AI is simultaneously overhyped and underhyped.
```
THE DUAL NATURE OF AI HYPE
┌───────────────────────────────┬───────────────────────────────┐
│ OVERHYPED ASPECT │ UNDERHYPED ASPECT │
├───────────────────────────────┼───────────────────────────────┤
│ • Human obsolescence in │ • Creation of novel, original │
│ 5 years │ insights │
│ • Passive reliance with no │ • Solutions outside existing │
│ creative purpose │ human corpus of knowledge │
└───────────────────────────────┴───────────────────────────────┘
```
Patel firmly rejects the overhyped notion that humans will become obsolete within five years, leaving everyone sitting idly on beaches while creative capacity disappears. Instead, jobs will be reconfigured as people adapt and add value in new ways. Conversely, what is underhyped is AI's capacity for generating original insights. Currently, models function primarily as aggregation mechanisms—sourcing and organizing the existing corpus of human knowledge. Moving forward, AI will generate net-new insights that do not exist anywhere in human history, fundamentally reimagining how society cures cancer, engineers new materials, and operates.
- **Core Shift:** Transitioning from historical scarcity-driven models to AI-enabled abundance.
- **Overhyped Threat:** The prediction of complete human obsolescence and joblessness within 5 years.
- **Underhyped Potential:** AI moving beyond knowledge aggregation to generating completely original, non-human insights.
---
[02:59](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=179s) The Second Phase of AI and Agentic Infrastructure Demands
Transitioning to agentic AI, Patel maps out the evolution of artificial intelligence into distinct phases. The first phase, spanning roughly two and a half years, centered on interactive chatbots capable of answering complex questions—a skill users have gradually mastered. Society is now entering the second phase: autonomous AI agents that perform tasks and jobs on behalf of humans. A current example is OpenAI's Deep Research feature, which operates autonomously for about 20 minutes to scour the web and return findings while a user gets coffee. As these autonomous operational windows expand from 20 minutes to hours, weeks, or months, the complexity of solvable problems will grow exponentially.
```
┌────────────────────────────────────────────────────────────────┐
│ THE TWO PHASES OF AI │
├───────────────────────────────┬────────────────────────────────┤
│ PHASE 1: CHATBOTS & PROMPTS │ PHASE 2: AUTONOMOUS AGENTS │
│ • User-driven Q&A │ • Computer use & task execution│
│ • 2-year adoption window │ • Operational for weeks/months │
└───────────────────────────────┴────────────────────────────────┘
```
However, scaling agentic AI creates two severe bottlenecks: infrastructure capacity and a trust deficit. On the infrastructure side, global power and compute limitations are forcing data centers to move directly to power sources rather than bringing power to data centers. With a future ratio where each person utilizes 10 to 100 autonomous AI agents, an 8-billion-person world will effectively demand the digital throughput capacity of 80 to 800 billion entities. While early adoption has centered on software coding and customer support, emerging "computer use" use cases—where agents navigate laptops like human workers—will place massive, exponential demands on network bandwidth and compute infrastructure. Simultaneously, deployment remains constrained by a trust deficit, as humans hesitate to delegate full authority to systems until safety and security are repeatedly proven.
- **Autonomous Horizon:** Execution windows for AI agents expanding from 20 minutes up to weeks or months.
- **Agent-to-Human Scale:** Future ratios estimated at 10 to 100 active agents per individual.
- **Global Population Throughput:** Effective human-agent workload equivalence scaling from 8 billion people up to 80 or 800 billion entities.
- **Infrastructure Dynamics:** Data center placement shifting toward power availability due to massive energy requirements of computer-use agents.
---
[06:35](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=395s) The Art of Prompting, Meta-Prompting, and Rapid Adoption
Egan steers the discussion toward the practical skills needed in daily AI interaction, asking about the art of prompting and meta-prompting. Patel notes that historical value was placed on accumulating knowledge, but in an era where intelligence is commoditized and freely available, competitive advantage relies on the nuance and quality of questions asked. Better questions allow people to target harder problems—and counterintuitively, harder problems carry a higher probability of success because they attract the best talent to collaborate.
To achieve step-function improvements in outputs, Patel advocates for "meta-prompting"—asking an AI model to write the ideal prompt for a given task before executing it. Demonstrating this technique on GPT-5 or similar models yields vastly superior, asymmetric improvements in generated content compared to raw human queries. Patel polls the audience on who has used meta-prompting, remarking on how rapidly humans are adopting these advanced techniques. This acceleration occurs because machines are learning human language rather than forcing humans to learn complex machine syntax.
- **Meta-Prompting Technique:** Instructing AI to generate an optimized prompt first, then using that output to perform the target task.
- **Problem Selection Dynamics:** Tackling harder, more complex problems raises success rates by attracting top-tier collaborators.
- **Adoption Speed:** Faster technological integration driven by machines learning human language rather than humans learning code.
---
[09:36](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=576s) Preserving Entry-Level Talent and Unlearning Experience
Addressing widespread economic anxieties, Egan points to warnings from Anthropic CEO Dario Amodei, who cautioned that AI could eliminate half of all entry-level white-collar jobs and push unemployment to 10–20% within one to five years. While acknowledging his friendship with Amodei and Cisco's investment in Anthropic, Patel strongly disagrees with eradicating entry-level positions, calling such strategy "the stupidest thing a company can do in the long term."
Patel highlights a fundamental contrast in how different generations adopt tools: a 28-year-old worker typically uses AI transactionally to automate existing tasks, whereas a 20-year-old worker treats AI as an intuitive sidekick, companion, and brainstorming partner. He illustrates this generational shift through an anecdote about his daughter, who used an iPad as early as age two or three; when she later encountered a physical newspaper, she kept tapping on the printed photos, bewildered that they wouldn't zoom in.
Eliminating early-career roles strips organizations of unconditioned perspectives. While experience is valuable, it often acts as a liability when past failures lead workers to dismiss ideas whose underlying conditions have changed. Entry-level hires force organizations to unlearn outdated assumptions. Patel shares that he actively mentors student interns and recent graduates specifically to absorb their fresh viewpoints, blending their untainted perspective with executive experience. Following a quick show of hands from the audience agreeing that entry-level jobs should not be eliminated, Patel stresses that productivity is the least interesting benefit of AI—the true value lies in solving completely novel problems.
- **Economic Forecast Debate:** Addressing predictions of 10–20% unemployment and 50% entry-level white-collar job loss within 1–5 years.
- **Strategic Imperative:** Rejection of corporate plans to eliminate entry-level positions to prevent losing fresh, unconditioned perspectives.
- **Personal Anecdote:** Patel's toddler daughter tapping physical newspaper photos expecting touchscreen zoom capabilities.
---
[15:02](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=902s) Navigating Social Transition and Emerging Security Risks
Acknowledging that technological shifts inevitably create friction, Egan asks about the societal growing pains accompanying AI adoption. Patel emphasizes that the technology community and public sector must partner to handle reskilling and retraining. Unlike tech workers accustomed to constant disruption, displaced physical workforce members—such as steel mill workers—cannot simply transition into AI prompt engineers overnight. Closing the skills gap through public-private education initiatives is critical to prevent severe societal disruption.
Beyond economic transitions, Patel warns of critical safety and security risks. AI tools are equally available to malicious adversaries, enabling nation-states and cybercriminals to execute hyper-sophisticated, bespoke attacks at machine scale. Cyber threats are rapidly evolving away from obvious, typo-ridden phishing emails from fictitious foreign princes toward highly targeted, convincing AI-driven operations.
- **Social Reskilling:** The necessity for joint public-private training initiatives to bridge the skills gap for workers outside the tech sector.
- **Evolving Threat Landscape:** Adversaries leveraging consumer-grade AI models to automate personalized, highly convincing cyberattacks.
<ElicitationsGroup message="Explore related topics and future discussions:">
<Elicitation label="Summarize Part 2 of this fireside chat" query="Provide a detailed walkthrough for Part 2 of the Cisco and CNN AI fireside chat."/>
<Elicitation label="Deep dive into agentic AI security challenges" query="What are the specific security frameworks Cisco uses to secure agentic AI ecosystems at machine scale?"/>
</ElicitationsGroup>
### [17:06](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=1026s) Securing AI Against Evolving Threats and Attacks
Jeetu Patel highlights the alarming sophistication of AI-driven social engineering attacks, describing a realistic scenario where an attacker might send a convincing message saying, "was great to see you at the our kids basketball game last night. Here are some pictures. Click to download." The moment the victim clicks, malware downloads onto the machine, initiating lateral movement across the network. To counter these and other risks, organizations must look beyond merely using artificial intelligence for cyber defense and actively focus on securing AI itself. Because large models are inherently susceptible to challenges like hallucinations, toxicity, and prompt injection attacks, robust mitigation mechanisms are essential.
Addressing these vulnerabilities requires a multi-layered security approach encompassing deep visibility into data inputs, rigorous model validation to ensure expected behavior, and robust runtime enforcement guardrails. Furthermore, autonomous agents require distinct identities that enable them to securely interact with other agents while maintaining strict, fine-grained control over their actions. For instance, an agent might be trusted to "craft an email on my behalf", but restrictions should prevent it from sending messages to "my boss and my board of directors", while still allowing it to communicate with "my team". Ultimately, navigating and solving these critical risk mitigation challenges on safety and security is a shared imperative for collective society as the agentic era unfolds.
### [18:25](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=1105s) Conclusion and Final Insights
Matt Egan wraps up the fireside chat by expressing deep appreciation for the discussion, thanking Jeetu Patel for sharing his expertise and valuable perspectives. Jeetu responds with gratitude, closing the session on a warm and positive note.
- Matt Egan expresses gratitude to Jeetu Patel for providing valuable insights during the discussion.
- Jeetu Patel responds warmly, concluding the conversation with words of appreciation.
## Notable quotes
> [00:04](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=4s) managing limited resources, which in turn drove invention. AI, however, introduces the possibility of true abundance — Jeetu Patel
> [00:04](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=4s) human obsolescence within five years, leaving everyone sitting idly on beaches while creative capacity disappears — Jeetu Patel
> [00:04](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=4s) generate net-new insights that do not exist anywhere in human history, fundamentally reimagining how society cures cancer, engineers new materials, and operates — Jeetu Patel
> [02:59](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=179s) each person utilizes 10 to 100 autonomous AI agents, an 8-billion-person world will effectively demand the digital throughput capacity of 80 to 800 billion entities — Jeetu Patel
> [06:35](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=395s) asking an AI model to write the ideal prompt for a given task before executing it — Jeetu Patel
> [06:35](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=395s) machines are learning human language rather than forcing humans to learn complex machine syntax — Jeetu Patel
> [09:36](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=576s) the stupidest thing a company can do in the long term — Jeetu Patel
> [09:36](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=576s) productivity is the least interesting benefit of AI—the true value lies in solving completely novel problems — Jeetu Patel
> [15:02](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=902s) steel mill workers—cannot simply transition into AI prompt engineers overnight — Jeetu Patel
> [17:06](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=1026s) was great to see you at the our kids basketball game last night. Here are some pictures. Click to download.
> [17:06](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=1026s) craft an email on my behalf
## Mentioned
Matt Egan, Jeetu Patel, Cisco, CNN, OpenAI, Deep Research, GPT-5, Anthropic, Dario Amodei
## Video description
Jeetu Patel, President and Chief Product Officer, Cisco
Matt Egan, Senior Reporter, CNN
Abstracts
Join Jeetu Patel, President and Chief Product Officer at Cisco, for a fireside chat with Matt Egan, award-winning CNN Senior Reporter, where they will explore the transformative power of agentic AI. They will discuss how organizations that effectively harness agentic AI will set themselves apart in the near future, while those that don't risk falling behind. Together, Jeetu and Matt will address the evolving landscape of safety, security, identity, and human collaboration with autonomous AI agents. Jeetu will also share Cisco's groundbreaking innovations that secure agentic AI ecosystems at machine scale, turning security from a challenge into a strategic advantage that accelerates AI adoption. Discover how these advancements empower organizations to lead with trust, agility, and resilience in the emerging agentic era.
思科Patel驳斥AI裁员论，警示网络瓶颈
思科总裁Jeetu Patel警告称，取消初级岗位会威胁组织的韧性，而自主AI智能体很快将使现有的能源和网络基础设施不堪重负。
思科的Jeetu Patel与CNN的Matt Egan探讨了从交互式聊天机器人向自主智能体AI的转变，届时AI或将以每人支配10到100个智能体的规模运行。Patel驳斥了人类将完全被淘汰的说法，并反对取消初级岗位，强调了未受思维定势限制的新鲜视角以及元提示（meta-prompting）技术的重要性。对话还涉及操作计算机的智能体对电力和网络基础设施提出的严苛要求，以及通过公私部门协同开展技能再培训和部署多层防御以应对AI网络威胁的必要性。
Cisco 总裁兼首席产品官 Jeetu Patel 指出，AI 正推动人类从资源匮乏走向潜在的富足时代；关于人类将在五年内完全被取代的说法被过度夸大，但 AI 产生全新原创见解的能力却被严重低估。 (0:04)
AI 正在从交互式聊天机器人迈向自主智能体（Agentic AI）阶段，未来人均可能使用 10 至 100 个智能体协同工作，这将使全球网络吞吐面临巨大考验，并受限于电力、算力短缺及人类信任赤字。 (2:59)
针对 AI 导致白领岗位流失的悲观预测，Patel 强烈反对取消初级岗位，强调缺乏固有成见的年轻员工对于打破组织陈旧假设至关重要。 (9:36)
应对智能体时代的风险不仅要使用 AI 进行网络防御，更需要防范针对 AI 自身的幻觉、提示词注入与机器级网络攻击，并为自主智能体构建细粒度的身份与操作权限控制。 (17:06)
Reading summary
Cisco President and Chief Product Officer Jeetu Patel argues that artificial intelligence marks a historical shift from scarcity management to societal abundance, driven not by basic task automation but by the capacity to generate entirely novel insights. However, realizing this agentic future depends on overcoming acute infrastructure bottlenecks, retaining junior talent to challenge legacy assumptions, and securing autonomous systems against machine-scale threats.
From Scarcity to Abundance and Original Insights (0:04)
Patel contends that while human history has been governed by scarcity and the management of finite resources, AI introduces the potential for genuine abundance to solve previously intractable systemic challenges. He frames current public perception as split between overhype and underhype. The overhyped scenario envisions total human obsolescence and mass joblessness within five years, which Patel rejects, asserting that work roles will reconfigure rather than disappear. Conversely, what remains underhyped is AI's evolution from an aggregation mechanism—synthesizing the existing corpus of human knowledge—into a generator of net-new insights capable of discovering novel cancer therapies and engineering new materials without historical precedent.
Autonomous Agent Expansion and Infrastructure Bottlenecks (2:59)
Patel traces AI's trajectory from an initial two-and-a-half-year phase of interactive chatbots to a second phase centered on autonomous agents executing complex workflows on behalf of humans. While current implementations such as OpenAI's Deep Research operate independently for roughly 20 minutes, Patel anticipates operational windows extending to weeks or months. With individuals projected to oversee between 10 and 100 autonomous agents, an eight-billion-person world could require the digital throughput equivalent of 80 to 800 billion entities.
This rapid expansion, particularly for "computer use" agents directly navigating operating systems, generates acute constraints. Global compute and power limitations are already compelling data centers to relocate to available power sources rather than bringing power to facilities. Deployment is further constrained by a societal trust deficit, as users remain cautious about delegating full operational authority to autonomous systems without proven safety track records.
Inquiry Quality and the Mechanics of Meta-Prompting (6:35)
As baseline intelligence becomes commoditized and broadly accessible, Patel argues that competitive advantage shifts from accumulating knowledge to mastering the nuance and precision of inquiries. Targeting harder problems counterintuitively increases the probability of success because difficult objectives attract premier talent. To maximize model performance, Patel highlights "meta-prompting"—instructing an advanced model like GPT-5 to compose the optimal prompt for an objective before executing it. He attributes the rapid enterprise adoption of such techniques to the fact that machines are learning natural human language rather than forcing operators to learn specialized programming syntax.
The Strategic Necessity of Entry-Level Roles (9:36)
Addressing warnings from Anthropic CEO Dario Amodei that AI could eliminate half of entry-level white-collar jobs and drive unemployment to 10–20% within one to five years, Patel firmly rejects corporate strategies that eliminate junior positions. While acknowledging Cisco's investment in Anthropic, Patel argues that institutional experience often acts as a liability when senior employees dismiss viable ideas because past conditions led to failure. Entry-level hires provide unconditioned perspectives essential for forcing organizations to unlearn outdated assumptions.
Patel observes a distinct generational divergence in adoption: a 28-year-old worker typically employs AI transactionally to automate established tasks, whereas a 20-year-old approaches AI intuitively as a creative brainstorming partner. He illustrates this conditioned divide through an anecdote of his young daughter expecting a physical newspaper photo to zoom upon touch, reinforcing that productivity gains are the least consequential aspect of AI compared to solving fundamentally new problems.
Societal Reskilling and Multi-Layered Agent Security (15:02)
To mitigate the friction of economic transition, Patel emphasizes that tech firms and public-sector leaders must coordinate workforce retraining, noting that displaced industrial personnel, such as steel mill workers, cannot instantly convert into prompt engineers. Simultaneously, malicious actors and nation-states are leveraging consumer-grade AI models to execute hyper-personalized, machine-scale social engineering attacks, moving past crude phishing schemes to bespoke communications that deploy malware via social pretexts.
Countering these risks requires moving beyond standard cyber defenses to securing AI architectures directly against hallucinations, data toxicity, and prompt injections. Patel advocates for a multi-layered security model combining deep input visibility, runtime guardrails, rigorous model validation, and dedicated agent identities. Under this governance model, organizations must implement granular operational constraints—such as permitting an autonomous agent to draft internal team updates while strictly barring it from dispatching communications to executive boards.
While Patel projects that agentic AI can unlock unprecedented societal abundance, realization of this trajectory remains conditional on solving severe power and compute bottlenecks, establishing identity and permission boundaries for autonomous agents, and executing structured public-private retraining to prevent workforce alienation.
精炼阅读摘要
Cisco 总裁兼首席产品官 Jeetu Patel 指出，人工智能正从基于知识聚合的交互式对话工具，跨入能够自主执行复杂任务的智能体（Agentic AI）时代；这一演进虽有望打破人类历史上的资源稀缺格局并生成前所未有的原创洞察，但其全面落地正面临基础设施能耗极限、初级人才战略误区以及高拟真网络安全威胁等多重严峻制约。
认知范式转向：从知识聚合到原创洞察 (0:04)
Patel 认为，人类文明史始终由资源稀缺性主导，而 AI 的终极潜能在于开启真正的资源充裕时代，为攻克癌症及新材料研发等历史顽疾提供全新路径。他明确反驳了所谓“人类将在五年内被完全淘汰且丧失创造性目标”的过度炒作言论，指出技术变革将重构而非彻底抹杀就业形态，人类社会将通过适应新场景继续创造独特价值。
与此同时，当前公众严重低估了 AI 产生全新原创洞察的能力。目前的模型主要充当人类既有知识体系的检索与聚合载体，而在未来的发展阶段，系统将能够跨越人类现有认知边界，自主生成历史上从未出现过的净新增洞察（net-new insights），从而从根本上重塑科研与产业运作机制。
智能体规模化演进与基础设施瓶颈 (2:59)
回顾技术演进，Patel 将过去约两年半的发展界定为第一阶段，即以用户驱动、问答为主的对话式工具；目前产业正迈入第二阶段，即代表人类自主执行任务的 AI 智能体阶段。目前如 OpenAI 的 Deep Research 功能已能自主开展约 20 分钟的网络搜寻与分析，未来智能体的独立作业周期预计将拓展至数小时、数周乃至数月，所能解决的任务复杂度也将呈指数级上升。
然而，智能体的大规模落地正遭遇严峻的基础设施与信任双重瓶颈。Patel 预测，未来每人可能拥有 10 至 100 个活跃的自主智能体，这意味着全球 80 亿人口将折合产生相当于 800 亿至 8000 亿个数字实体的吞吐量需求。伴随智能体接管类似人类操作电脑的“计算机使用”（computer use）场景，巨大的网络带宽和算力负荷正迫使数据中心从“引电就机”转向直接向电力能源富集地迁移；此外，在系统的安全性得到反复确证前，用户对全权委托系统运作仍存有显著的信任赤字。
提问能力重构与保留初级岗位的战略考量 (9:36)
在智能供给高度商品化的背景下，个人的竞争优势已从单纯累积既有知识转向提出高质量、深层次问题的能力。Patel 指出，解决越困难的问题往往越容易借助顶尖人才的协作取得成功；在实践中，通过“元提示”（Meta-Prompting，即让 AI 在执行前先构建最佳提示词）能够带来跨越式的产出提升。人机交互正因为机器逐步理解人类自然语言而加速普及，摆脱了以往人类必须学习机器语法的门槛。
针对 Anthropic 首席执行官 Dario Amodei 提出的“AI 可能在 1 至 5 年内消除 50% 白领初级岗位并将失业率推高至 10%–20%”的行业预测，Patel 虽认可两家机构的合作关系，但坚决反对企业削减初级职位的做法，直言这是长期维度上最愚蠢的企业决策。他指出，年长员工的丰富经验有时会因受挫记忆而演变成拒绝新机遇的思维包袱，而年轻一代将 AI 视作日常头脑风暴伙伴，能带来不受固有成见束缚的全新视角。保留初级人才不仅能促使企业主动打破陈旧经验（unlearn），也是挖掘 AI 探索全新问题而非单纯追求生产力提升的组织基石。
社会技能转型阵痛与智能体多层级安全防御 (15:02)
面对技术红利伴随的社会结构性摩擦，Patel 强调公共部门必须与科技产业紧密协作，共同推进劳动力的再培训。与习惯技术迭代的从业者不同，钢铁厂工人等传统物理实体产业劳动力无法瞬间转变为提示词工程师，若缺乏强有力的跨部门教育支持，技能断层将引发严重的社会动荡。
在安全合规维度，攻击者正利用同等水平的 AI 工具发动高度定制化、难以察觉的社会工程学攻击，以往粗糙的钓鱼邮件已被利用个人社交细节精准编造的恶意指令所取代。防御策略必须从“借助 AI 实施网络防御”转向“保卫 AI 系统自身安全”，建立涵盖数据输入深度可见性、模型验证以及运行拦截的多层防护体系，以应对幻觉、毒性内容和提示注入。此外，自主智能体必须被赋予独立的数字身份与细粒度权限控制机制，防止其在跨团队交互或代表用户行动时出现未经授权的越权扩散。
本场对话的核心结论表明，智能体革命的核心价值并非止步于常规的生产效率提升，而在于拓展人类解决未知难题的边界。然而，如何在面对潜在高达数百倍于人口规模的数字实体时解决全球能源网络瓶颈、如何化解非技术劳动力的结构性失业隐忧，以及如何在缺乏细粒度安全与身份验证机制的现状下建立对自主系统的可验证信任，依然是智能体生态全面铺开前亟待社会各界协同解答的根本问题。
Detailed notes — zh
## Background
本内容由 Ai4 于 2025 年 8 月 24 日发布在 YouTube 平台上。视频记录了思科（Cisco）总裁兼首席产品官 Jeetu Patel 与 CNN 资深记者 Matt Egan 展开的炉边对话，重点探讨智能体 AI（Agentic AI）的变革力量、组织在安全与信任层面的应对策略，以及思科在保护机器级 AI 生态方面的安全创新。
## TL;DR
对话深入探讨了 AI 如何推动人类从“资源匮乏”走向“资源丰裕”的新范式，指出当前外界高估了人类短期被淘汰的风险，却低估了 AI 产生原创性洞见的能力。随着自主运行的 Agent 时代到来，电力算力短缺与信任赤字成为核心瓶颈，而提问能力和元提示（Meta Prompting）正在成为人机协作的关键壁垒。面对就业冲击，Patel 坚决反对淘汰初级岗位，强调新生代的数字化直觉与“主动遗忘”经验负债的价值；同时呼吁全社会重视技能重塑，并构建涵盖数据可见性、独立机器身份和细粒度权限控制的 AI 本身安全防护体系。
## Atmosphere & style
整场对话节奏沉稳而富有洞察力，展现出科技高管与资深财经媒体人之间高密度的思想碰撞。Patel 的表达兼具宏观愿景与生活化案例，既有对技术指数级爆发的由衷兴奋，也透露出对蓝领转型阵痛与网络安全威胁的清醒克制与现实关怀。现场互动真诚自然，在理性剖析技术演进的同时，传递出一种立足于安全防线、拥抱代际创新的建设性乐观基调。
## Key takeaways
- [00:04](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=4s) AI 正在推动人类社会从以“资源匮乏”为核心驱动的发明与分配模式，迈向以“资源丰裕”为可能性的全新解题范式。
- [00:04](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=4s) 行业过度高估了“人类将在 5 年内被彻底淘汰并失去创造力”的极端悲观论调，工作形态虽会被重构，但人类不会全面过时。
- [00:04](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=4s) 当前大模型本质上是既有知识库的聚合机制，而未来 AI 真正被低估的潜力在于能够跨越现有知识边界产生前所未有的“原创性洞见”。
- [02:59](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=179s) 智能体 AI 正在经历从对话式问答向长周期（数小时、数天乃至数月）自主执行复杂任务的指数级跨越。
- [02:59](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=179s) 若未来每人配备 10 至 100 个自主 Agent，全球产能吞吐将呈几何级暴增，从而使电力和算力等基础设施面临硬性约束。
- [02:59](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=179s) 人类目前赋予 AI 的实际权限远落后于其真实能力，安全感与确定性不足导致的“信任赤字”是放权 Agent 的核心瓶颈。
- [06:35](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=395s) 在知识高度充沛与商品化的时代，竞争壁垒已从知识储备转移到“提问的精妙度与质量”。
- [06:35](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=395s) 采用“元提示”（Meta Prompting）让 AI 先设计优质提示词再执行任务，能够带来阶跃式的生成效果提升。
- [06:35](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=395s) 自然语言成为机器主动理解人类的通用接口，极大压缩了人类采纳并掌握前沿新技术的周期。
- [09:36](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=576s) 企业若因 AI 普及而淘汰所有初级岗位将是重大失误，年轻一代具备原生的数字直觉和将 AI 视作实时副驾的协作模式。
- [09:36](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=576s) 经验往往会成为阻碍创新的负债，资深员工需要“主动遗忘”（Unlearn），将行业积淀与年轻人的全新破局视角相结合。
- [15:02](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=902s) 科技行业必须承担社会责任，不能沉溺于乐观泡泡，应与公共部门协作推进全社会的技能重塑（Reskilling）以弥合转型阵痛。
- [15:02](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=902s) 攻击者与防御者享有同等水平的 AI 工具，网络威胁正从粗制滥造的钓鱼演进为基于个体特征量身定制的高级社交工程攻击。
- [17:06](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=1026s) 网络防御的重心必须从单纯“利用 AI 进行防御”转向“保护 AI 本身的安全”，以化解幻觉、毒性输出及提示词注入等内生风险。
- [17:06](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=1026s) 构建 Agent 生态安全屏障需要实现端到端数据可见性、模型行为验证、运行时强制护栏以及针对 Agent 的独立身份与细粒度权限控制。
## Detailed walkthrough
### [00:04](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=4s) AI 革命的本质：从资源匮乏到丰裕，以及被高估与低估的面向
CNN 资深记者 Matt Egan 以思科（Cisco）执行副总裁兼首席产品官 Jeetu Patel 曾经的一句论断开场——“AI 是我们一生中见过最重大的变革”。Egan 随即提问，置身于这场前所未有的浪潮中，人类社会和世界长期以来的哪些底层假设必须被重新审视。Patel 坦言，当前身处科技行业令人无比兴奋，而人类有史以来所有文明与发明创造几乎都建立在“资源匮乏”（Scarcity）这一核心假设之上，匮乏不仅倒逼人们精打细算地平衡资源，也在某种程度上充当了“发明之母”。然而，AI 正在将人类带入一个真正可能实现“资源丰裕”（Abundance）的新纪元，彻底打开过去无法企及的解题空间。尽管这一目标不可能在明后年一蹴而就，但其带来的机遇将是颠覆性的。
谈及外界对 AI 的认知偏差，Patel 指出当前行业存在一个被严重“高估”的论调和一个被严重“低估”的潜力。被过度高估的，是所谓“人类将在五年内被彻底淘汰”的悲观论断。Patel 坚决驳斥了“人类届时无事可做、只能躺在海滩上、创造力完全枯竭”的设想，强调人类历史上每一次变革都重新找到了创造价值的方式，工作形态固然会重构，但绝不会导致人类全面过时。相反，真正被低估的则是 AI 带来“独创性洞见”（Original Insights）的能力。当前的以大语言模型（LLM）为主的 AI 本质上仍是一个“聚合机制”（Aggregation Mechanism），擅长检索并梳理人类现有的知识库；但在未来，AI 将能够跨越现有知识库的边界，产生人类前所未见的全新原创见解，从而从根本上重塑攻克癌症、研发新材料乃至运行整个社会的方式。
- **核心假设变迁**：从以“匮乏”为驱动的资源分配模式，迈向以“丰裕”为可能性的解题新范式。
- **被过度高估的威胁**：断言人类在 5 年内成为无用阶级、失去所有创造力并沦为废人的极端言论。
- **被严重低估的潜力**：超越对既有人类知识库的汇总聚合，AI 将诞生出人类从未触及过的“原创性洞见”。
---
### [02:59](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=179s) 智能体 AI（Agentic AI）时代的到来与基础设施与信任的双重瓶颈
讨论切入智能体 AI（Agentic AI）的未来蓝图。Patel 将 AI 的演进划分为清晰的阶段：第一阶段是过去两年半里大家所熟知的“智能聊天机器人”，它们能以看似神奇的方式回答复杂问题；而当前正全面迈入的第二阶段，则是能够代表人类自主执行复杂任务和端到端工作的“AI 智能体”（Agents）。以目前的深度研究工具（Deep Research）为例，智能体已经可以在人类喝一杯咖啡的 20 分钟内自主检索全网并汇总报告。Patel 强调，当这种自主执行的时间窗口从 20 分钟延伸到 2 小时、2 天、2 周乃至 2 个月时，AI 解决复杂问题的能力将呈现爆发式的指数级增长。
```mermaid
flowchart TD
accTitle: Agentic AI Evolution and System Constraints
subgraph Progression["AI 智能体自主工作时长演进"]
A["对话式问答<br/>（第一阶段 Chatbots）"] --> B["20 分钟自主研究<br/>（如 Deep Research）"]
B --> C["数小时至数天<br/>（复杂流程自主执行）"]
C --> D["数周至数月<br/>（端到端重大问题攻关）"]
end
subgraph Expansion["产能乘数效应"]
D --> E["80 亿人口 × (10~100 个智能体)"]
E --> F["相当于 800 亿~8000 亿人吞吐产能"]
end
subgraph Bottlenecks["核心制约瓶颈"]
G["基础设施约束<br/>（电力短缺、算力与网络受限）"]
H["信任赤字（Trust Deficit）<br/>（安全、隐私与权限开放顾虑）"]
end
Progression -. 触发.-> Bottlenecks
Expansion -. 加剧.-> Bottlenecks
```
然而，这种指数级增长随即带来了两大严峻瓶颈：
1. **基础设施硬约束（Infrastructure Constraints）**：全球电力与算力严重匮乏。如今数据中心不再等电力输送过来，而是“哪里有电，数据中心就建到哪里”。如果未来每个自然人配备 10 到 100 个全天候工作的自主智能体，全球 80 亿人口的实际产能吞吐将相当于 800 亿甚至 8000 亿人。尤其是当“计算机接管操作”（Computer Use，即 AI 模拟人类在电脑屏幕上直接操作软件完成任务）普及后，对算力和网络带宽的消耗将呈几何级暴增。
2. **信任赤字（Trust Deficit）**：人类目前赋予 AI 的实际权限远低于 AI 当前的真实能力边界。由于缺乏足够的安全感与确定性，用户与企业不敢轻易放权。这种信任必须随着时间推移、伴随完善的安全性（Safety）与安全防护（Security）机制被逐步建立。
---
### [06:35](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=395s) 提问的艺术与元提示（Meta Prompting）
针对人机协作的具体实践，Patel 深入剖析了提示词工程（Prompting）与提问能力的底层逻辑。他结合自身养育 14 岁女儿的经历指出，过去几代人的核心信念是“知识积累决定竞争力”，但在智能与知识日益充沛、廉价甚至高度商品化的今天，真正的分水岭已经转移到“提问的精妙度与质量”。提出问题的质量直接决定了所能解决问题的难度，而解决最难的问题反而拥有最高的成功概率——因为最具挑战性的难题往往能吸引最顶尖的人才，而商业本质上是一场“团队运动”。
在具体操作层面，Patel 介绍了随着 GPT-5 等新一代模型发展而愈发关键的技术——“元提示”（Meta Prompting）。元提示的核心在于：不要直接让 AI 执行最终任务，而是让 AI 为你设计一段最优质的提示词，再用该提示词去生成最终结果。Patel 现场举例：如果你需要生成一张特定场景的图像，先让 AI 构思一套专门用于生成该图像的详细 Prompt，然后将这段元提示复制回去执行，产出的画面质量与直接提问相比会呈现出“不对称的阶跃式提升”。当 Patel 询问现场有多少人使用过元提示时，看到许多与会者举手，他感叹人类吸收新技术的周期正在被极度压缩，根本原因在于“如今是机器在主动学习人类的自然语言，而非人类去苦学机器语言”。
- **知识价值转移**：单纯的知识记忆与信息储备正在快速商品化，提问的深度与思考维度成为核心壁垒。
- **元提示操作法则**：通过“让 AI 指导人类如何向 AI 提问”，利用模型的自我反思与丰富表征获得跃迁级别的输出效果。
- **人机交互质变**：自然语言成为通用接口，使技术扩散与采纳速度产生了前所未有的时间压缩效应。
---
### [09:36](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=576s) 拒绝淘汰初级员工：经验的负债与新生代视角的不可替代性
话题转向 AI 对就业市场的冲击。Matt Egan 提及 Anthropic 首席执行官 Dario Amodei 的公开警告——AI 可能在未来 1 到 5 年内消灭一半的初级白领工作，并将失业率推高至 10% 到 20%。对此，Patel 表达了不同看法。他首先肯定了 Dario 的卓越成就（思科也是 Anthropic 的投资方），但坚决反对“企业应淘汰所有初级岗位（Entry-level jobs）”的观点，并直言这是“一家公司从长远来看能做出的最愚蠢的决定”。
为了阐释这一观点，Patel 对比了不同代际使用 AI 的行为差异：
- **28 岁的职场人**：往往以高度“任务导向”和“事务性”（Transactional）的方式使用 AI，将其视作查答案或自动化既定工作流的工具。
- **20 岁的年轻群体**：则天然将 AI 当作实时协作的“副驾”（Sidekick）与头脑风暴伙伴（Brainstorming Partner）。
Patel 分享了一个生动的生活细节：他的女儿两三岁时便接触 iPad，习惯了触摸缩放；当她第一次见到传统纸质报纸时，本能地用手指去不停点击报纸上的印刷照片，困惑于图片为什么不能放大。这种原生的数字化直觉正是企业必须注入的新鲜血液。
Patel 深刻指出，“经验在很多时候会变成沉重的负债”。职场中资深人士最常说的一句话往往是“五年前我们试过这个方案，根本行不通”，却忽略了五年间周围五万个底层假设可能早已发生巨变。老旧经验会本能地阻碍创新，因此资深员工必须学会“主动遗忘”（Unlearn）。年轻一代不受过往经验的束缚，能带来全新的破局视角，而这种独特视角与资深人员的行业经验结合时，才会产生真正的奇迹。Patel 补充道，AI 最迷人的地方绝非单纯提升生产效率，而是赋能人类解决以前由于知识库视角缺失而完全不敢想象的全新难题。
---
### [15:02](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=902s) 转型阵痛、社会重塑与日益复杂的安全威胁
在回应现场关于技术转型阵痛的提问时，Patel 承认任何颠覆性技术普及都会伴随阵痛。他特别提醒科技界同仁跳出行业固有的“技术乐观泡泡”：“在科技圈，我们总觉得颠覆是家常便饭；但当一名钢铁厂工人面临岗位淘汰时，他们不可能一转眼就变成一名 AI 提示词工程师。”科技行业必须承担起社会责任，与公共部门紧密合作，推进全社会的技能重塑（Reskilling）与再培训计划，尽全力弥合教育与技能鸿沟。
除了社会层面的就业重塑，Patel 将另一大核心危机聚焦于**安全与防护（Safety and Security）**。防御者所使用的前沿 AI 工具，攻击者和恶意敌对势力（包括国家级黑客组织）同样垂手可得。未来的网络攻击将变得极其复杂且高度定制化。过去的钓鱼攻击往往粗制滥造——例如邮件里满是错别字、自称来自某异域国家的王子请求提供银行账户以便转账 1000 万美元；而在 AI 时代，攻击手段已经升级为天衣无缝、针对个体行为特征量身定制的深度伪造与精准欺诈，若缺乏强有力的安全机制加以遏制，其后果将是灾难性的。
### [17:06](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=1026s) 保障 AI 自身安全：从新型威胁到细粒度 Agent 权限控制
Jeetu Patel 举例阐述了新型网络攻击如何利用极度个性化的社交工程手段发起渗透。攻击者可能会伪造一封看似极为亲近的邮件：“昨晚在孩子的篮球赛上见到你真高兴，这里附上几张照片，点击即可下载”（"was great to see you at the our kids basketball game last night. Here are some pictures. Click to download."）。受害者一旦点击，恶意软件便会立即下载到设备中，并在内网系统展开横向移动（lateral movement）。Patel 强调，面对这种威胁，企业与行业绝不能仅停留在“利用 AI 进行网络防御”（using AI for cyber defense）的传统层面上，更核心的紧迫任务是“保护 AI 本身的安全”（secure AI itself）。因为底层大模型自身就存在幻觉（hallucination）、毒性有害内容（toxicity）以及极易遭受提示词注入攻击（prompt injection attacks）等固有风险，必须从机制上对其进行有效化解。
要构建稳固的 Agentic AI 安全屏障，必须在数据、模型与运行治理层面建立全方位的管控机制。首先要实现全方位的数据可见性，明确掌握哪些数据被输入到了模型中；其次需进行严密且持续的模型验证，确保模型按预期逻辑运行；同时必须在执行层面部署强制性的运行时安全护栏（runtime enforcement guardrails）。在 Agent 相互协作的生态中，AI Agent 还必须拥有专门的、可验证的独立机器身份（identity），并辅以严格的细粒度权限控制（fine-grained control）。Patel 用日常协作场景形象地说明：“你或许允许一个 Agent 代表你起草并发送邮件给团队成员，但绝不会允许它直接把邮件发给你的老板或董事会成员。”这种安全防护与风险缓释机制，是整个社会在拥抱自主智能体时代必须共同攻克的核心议题。对话最后，Matt Egan 对 Patel 分享的前沿洞见表达了高度感谢，双方在对 Agent 时代安全发展方向的共识中结束了这场交流。
```mermaid
flowchart TD
accTitle: AI 安全与 Agent 治理防护架构
subgraph Governance ["模型与数据治理"]
A["数据输入可见性<br/>(Visibility)"] --> B["模型行为验证<br/>(Validation)"]
end
subgraph ExecutionSecurity ["运行时与 Agent 身份防护"]
C["运行时安全护栏<br/>(Runtime Guardrails)"]
D["Agent 独立身份验证<br/>(Identity Verification)"]
E["细粒度权限控制<br/>(Fine-Grained Control)"]
end
Governance --> ExecutionSecurity
ExecutionSecurity --> F["受控任务执行<br/>(区分团队普通操作与高管/董事会高危操作)"]
```
- **保护 AI 本身（Securing AI Itself）面临的内生风险**：包括模型幻觉、有害输出（Toxicity）以及对抗性的提示词注入攻击（Prompt Injection Attacks）。
- **Agent 生态安全防护的关键支柱**：
- **数据与模型可见性**：端到端审查输入数据流并进行模型预期行为验证。
- **运行时强制防护栏**：在 Agent 执行操作时进行动态安全策略拦截与合规检查。
- **Agent 机器身份与细粒度权限**：赋予 Agent 可信身份认证，严格划定操作边界，防止未经授权的敏感指令扩散。
## Notable quotes
> [00:04](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=4s) AI 是我们一生中见过最重大的变革 — Jeetu Patel
> [00:04](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=4s) 人类届时无事可做、只能躺在海滩上、创造力完全枯竭 — Jeetu Patel
> [02:59](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=179s) 哪里有电，数据中心就建到哪里 — Jeetu Patel
> [06:35](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=395s) 如今是机器在主动学习人类的自然语言，而非人类去苦学机器语言 — Jeetu Patel
> [09:36](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=576s) 一家公司从长远来看能做出的最愚蠢的决定 — Jeetu Patel
> [09:36](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=576s) 五年前我们试过这个方案，根本行不通 — Jeetu Patel
> [15:02](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=902s) 在科技圈，我们总觉得颠覆是家常便饭；但当一名钢铁厂工人面临岗位淘汰时，他们不可能一转眼就变成一名 AI 提示词工程师 — Jeetu Patel
> [17:06](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=1026s) was great to see you at the our kids basketball game last night. Here are some pictures. Click to download. — Jeetu Patel
> [17:06](https://www.youtube.com/watch?v=YSIg7qfdi5I&t=1026s) 你或许允许一个 Agent 代表你起草并发送邮件给团队成员，但绝不会允许它直接把邮件发给你的老板或董事会成员 — Jeetu Patel
## Mentioned
Jeetu Patel, Matt Egan, 思科 (Cisco), CNN, Anthropic, Dario Amodei, GPT-5, Deep Research, iPad
## Video description
Jeetu Patel, President and Chief Product Officer, Cisco
Matt Egan, Senior Reporter, CNN
Abstracts
Join Jeetu Patel, President and Chief Product Officer at Cisco, for a fireside chat with Matt Egan, award-winning CNN Senior Reporter, where they will explore the transformative power of agentic AI. They will discuss how organizations that effectively harness agentic AI will set themselves apart in the near future, while those that don't risk falling behind. Together, Jeetu and Matt will address the evolving landscape of safety, security, identity, and human collaboration with autonomous AI agents. Jeetu will also share Cisco's groundbreaking innovations that secure agentic AI ecosystems at machine scale, turning security from a challenge into a strategic advantage that accelerates AI adoption. Discover how these advancements empower organizations to lead with trust, agility, and resilience in the emerging agentic era.
[Source](https://www.youtube.com/watch?v=YSIg7qfdi5I)
Source quotes
[0:04] Jeetu Patel: AI 是我们一生中见过最重大的变革
[0:04] Jeetu Patel: 人类届时无事可做、只能躺在海滩上、创造力完全枯竭
[2:59] Jeetu Patel: 哪里有电，数据中心就建到哪里
[6:35] Jeetu Patel: 如今是机器在主动学习人类的自然语言，而非人类去苦学机器语言
[9:36] Jeetu Patel: 一家公司从长远来看能做出的最愚蠢的决定
[9:36] Jeetu Patel: 五年前我们试过这个方案，根本行不通
[15:02] Jeetu Patel: 在科技圈，我们总觉得颠覆是家常便饭；但当一名钢铁厂工人面临岗位淘汰时，他们不可能一转眼就变成一名 AI 提示词工程师
[17:06] Jeetu Patel: was great to see you at the our kids basketball game last night. Here are some pictures. Click to download.
[17:06] Jeetu Patel: 你或许允许一个 Agent 代表你起草并发送邮件给团队成员，但绝不会允许它直接把邮件发给你的老板或董事会成员
Source quotes
[0:04] Jeetu Patel: managing limited resources, which in turn drove invention. AI, however, introduces the possibility of true abundance
[0:04] Jeetu Patel: human obsolescence within five years, leaving everyone sitting idly on beaches while creative capacity disappears
[0:04] Jeetu Patel: generate net-new insights that do not exist anywhere in human history, fundamentally reimagining how society cures cancer, engineers new materials, and operates
[2:59] Jeetu Patel: each person utilizes 10 to 100 autonomous AI agents, an 8-billion-person world will effectively demand the digital throughput capacity of 80 to 800 billion entities
[6:35] Jeetu Patel: asking an AI model to write the ideal prompt for a given task before executing it
[6:35] Jeetu Patel: machines are learning human language rather than forcing humans to learn complex machine syntax
[9:36] Jeetu Patel: the stupidest thing a company can do in the long term
[9:36] Jeetu Patel: productivity is the least interesting benefit of AI—the true value lies in solving completely novel problems
[15:02] Jeetu Patel: steel mill workers—cannot simply transition into AI prompt engineers overnight
[17:06] was great to see you at the our kids basketball game last night. Here are some pictures. Click to download.
[17:06] craft an email on my behalf
Original transcript
[0:04] I'd love to start with your views on the
[0:07] AI revolution. You've said that AI is
[0:09] bigger than anything you've seen in your
[0:11] lifetime. Given that, are there any
[0:14] assumptions about the world and society
[0:17] that must be reconsidered to fit this
[0:22] new world that we're all entering?
[0:26] Look, I think firstly it's a really
[0:27] interesting time to be uh be alive in
[0:29] tech. I think we'd all agree. Um there's
[0:32] a uh um there's a core set of
[0:36] assumptions that we've all kind of lived
[0:38] with for as long as humans have been
[0:41] around which is around this notion of
[0:44] scarcity and that everything is around
[0:46] you know we have a scarce amount of
[0:48] resources. We got to balance those
[0:50] resources so that we can make sure that
[0:52] we um you know and scarcity has been the
[0:55] mother of invention to some degree. I
[0:58] think we are now entering a phase where
[1:03] abundance is actually possible and it
[1:06] might actually open up a whole set of
[1:08] possibilities of solving problems that
[1:11] we couldn't solve before. Uh I still
[1:13] think we're a ways away from it. that's
[1:14] not going to happen by next year, but I
[1:17] do feel like, you know, there's going to
[1:18] be a tremendous amount of opportunity
[1:21] for solving problems that we couldn't
[1:24] solve before. U and
[1:27] and I feel like there's one thing that's
[1:29] overhyped about AI and there's one thing
[1:31] that's underhyped about AI. Thing that's
[1:33] overhyped is I reject the notion that
[1:35] humans are going to be obsolete in like
[1:36] 5 years where we're not going to have
[1:38] anything to do and we're going to be
[1:39] sitting on the beach and our creative
[1:41] juices are just going to be completely
[1:43] obsolete. it it doesn't make any sense.
[1:45] I think we've always found a way to go
[1:47] out and add value. Now, all of our jobs
[1:49] will have some degree of
[1:51] reconfiguration, but it's not going to
[1:53] be where we we get to be be obsolete.
[1:55] So, I think that part is overhyped about
[1:57] AI. What's underhyped about AI is the
[2:01] original insights. Right now, AI is
[2:03] largely especially with large language
[2:05] models and all of that is largely an
[2:07] aggregation mechanism. It does a good
[2:09] job going out and sourcing the corpus of
[2:12] knowledge that humans have and being
[2:14] able to go figure out if there's
[2:16] insights that could be had from it. I
[2:18] think moving forward original insights
[2:21] will be created in ways that it's it's
[2:25] going to be very hard um to um go out
[2:29] and replicate and those original
[2:30] insights will actually not exist in the
[2:32] human corpus of knowledge and that'll
[2:34] allow us to solve problems that I don't
[2:36] think we've actually thought about
[2:37] solving yet.
[2:38] >> Wow. And and that in my mind is um is is
[2:41] a very exciting kind of possibility that
[2:42] we can the way that you solve cancer,
[2:44] the way that you actually make sure that
[2:46] you create materials, the way that um
[2:49] you know kind of um uh society operates,
[2:52] all of those things will need to get
[2:55] fundamentally reimagined. Uh and I think
[2:57] that's going to be cool.
[2:59] >> You know, let's talk about agentic AI a
[3:02] bit. Um, you've said that you can see a
[3:03] future where billions of AI agents are
[3:07] working together and in in harmony on on
[3:10] behalf of humans. How far are we from
[3:13] that future and and what will the
[3:15] consequences of that be?
[3:18] >> So, I think we're in this squarely in
[3:20] the second phase of AI, right? The first
[3:22] phase was these intelligent chat bots.
[3:23] You could answer questions and it felt
[3:25] like magic. Um, and you know, we are
[3:28] about two and a half years into that
[3:29] phase. most of us have now gotten to
[3:33] some level of dexterity of answer asking
[3:36] nuanced questions so that you can get
[3:38] good answers back. Uh we'll talk a
[3:40] little bit about that if we have some
[3:42] time. I think the second phase of AI is
[3:44] where you have these agents that can
[3:46] autonomously conduct tasks and jobs on
[3:48] behalf of humans. Now today, if you
[3:51] think about probably the one of the most
[3:53] effective agents that you have out
[3:54] there, it's deep research, right? Like
[3:57] you can it probably works autonomously
[3:59] for about 20 minutes. You can ask it to
[4:02] go do something. You can go get a cup of
[4:03] coffee, ask it to do some research,
[4:05] it'll come back to you. It'll go, you
[4:08] know, scour the web, come back to you
[4:09] with some answers. I think as that
[4:12] window expands and if you have from you
[4:15] know 20 minutes to two hours that an
[4:18] agent can work autonomously to to two
[4:21] days to two weeks to two months you can
[4:25] start seeing the exponentiality of the
[4:28] kind of problems that can be solved with
[4:30] AI. Uh it it'll be very cool. Now it'll
[4:33] also create a tremendous amount of
[4:35] constraints on the other side which is I
[4:38] think there's going to be a massive
[4:39] constraint on infrastructure.
[4:42] Um we don't have enough power in the
[4:44] world to go out and satiate the needs
[4:46] for AI today. That's why you're actually
[4:49] seeing data centers being built in every
[4:51] place where you might have power. Data
[4:53] centers are going to where power is
[4:55] rather than power coming to where data
[4:56] centers are. um we don't have enough
[4:59] compute capacity.
[5:01] And if you see these agents that are
[5:03] going to autonomously continue to keep
[5:05] working round the clock and if you're if
[5:09] the ratio is for every human you might
[5:11] have 10 to 100 agents,
[5:14] a world of 80 8 billion people might
[5:16] feel like we have the throughput
[5:18] capacity of maybe 80 or 800 billion
[5:20] people. Um but the the use case that'll
[5:24] completely change the assumptions of
[5:25] infrastructure is just starting to come
[5:28] up. So you know if you think about the
[5:30] use cases that have actually gotten
[5:31] volume in AI right now it's coding it's
[5:35] customer support all of those use cases.
[5:37] Computer use I think as a use case where
[5:40] an agent pretends to be a human on your
[5:43] laptop and actually conducts tasks for
[5:45] you. computer use as a use case will
[5:48] have a exponentiality to it on use of
[5:52] compute on use of the network. That's
[5:54] going to be pretty huge. And then the
[5:56] second big area that we should think
[5:58] about is there's a there's going to be a
[5:59] trust deficit where people just don't
[6:02] trust these systems right now. If they
[6:03] don't trust these systems, they're not
[6:05] going to use them. Safety and security
[6:06] is pretty important and those systems
[6:09] that that trust has to get earned over
[6:12] time by AI from humans in order for
[6:15] humans to actually let it in a little
[6:18] bit more. The capacity of what AI can do
[6:20] even today is far more than what humans
[6:23] are allowing it to do today. And I think
[6:27] the trust deficit is a big part of it.
[6:28] And so those are the two areas, the
[6:30] constraints on infrastructure and the
[6:32] trust deficit that'll have to get played
[6:33] out.
[6:35] So you mentioned prompting. Can you talk
[6:37] a little bit about the art of asking
[6:40] questions to AI systems uh and
[6:43] metaprompting because I think this is
[6:45] something that is is pretty um practical
[6:49] that would help our audience
[6:50] >> help everyone. Yeah. So um and you know
[6:53] Fay talked about this a little bit on
[6:57] if you think about the uh I have a
[6:59] 14-year-old daughter right and
[7:02] the assumption
[7:05] during my entire lifetime was that
[7:07] knowledge is really important. You need
[7:09] to make sure that you accumulate as much
[7:11] knowledge as possible and that's what's
[7:13] going to get you differentiated in the
[7:14] world. Where we are now is knowledge and
[7:17] intelligence is going to get to be in
[7:18] abundance and actually you know
[7:20] available for free and to some degree
[7:22] commoditized.
[7:24] But where the art comes in is the nuance
[7:28] with which you can ask questions
[7:30] right and the quality of questions that
[7:32] you ask will determine the quality of
[7:36] problems that you can solve. And the
[7:38] quality of problems that you solve
[7:40] actually is directly proportionate to
[7:42] the success of the outcome. The harder
[7:44] the problem,
[7:45] the higher the probability of success,
[7:47] which is completely counterintuitive.
[7:49] Why is that? Because typically the
[7:52] hardest problems are the ones that
[7:53] attract the best people. And business is
[7:55] a team sport. And so if you have the
[7:57] hardest problems, you'll actually get
[7:59] the best amount of, you know,
[8:01] probability for a successful outcome.
[8:03] So on the prompting side, what's
[8:06] actually now starting to happen with
[8:07] GPT5, you're starting to see this more
[8:10] is this notion of meta prompting
[8:12] where you can ask a question,
[8:15] but that prompt is asking AI to give you
[8:19] a prompt that's going to give you an
[8:21] answer for that prompt.
[8:23] And try this out sometimes, and it's
[8:25] actually quite remarkable. Ask it to go
[8:27] generate an image for you with a prompt.
[8:30] you know, I I wanted to see if um you
[8:32] know, whatever the image might be that
[8:34] you want to generate for whatever
[8:36] situation, ask AI to generate the image,
[8:39] then ask it not to generate the image,
[8:41] but to give you a prompt to generate the
[8:43] image. And then they will give you a
[8:45] prompt and you copy and paste that
[8:46] prompt and see the see the quality of
[8:49] image you're going to get from that
[8:50] metaprompt rather than the prompt that
[8:52] you might have provided it. And you'll
[8:54] actually see a stark difference and an
[8:56] asymmetry with which you can actually
[8:59] have this massive step function
[9:01] improvement in the output that you get.
[9:04] >> You know,
[9:04] >> that's that's really cool. I I can't
[9:06] wait.
[9:06] >> How many have done how many have used a
[9:07] metaprompt over here with a show of
[9:08] hands?
[9:09] >> Wow.
[9:10] >> Great. This this is the other thing
[9:12] that's crazy is um the the rate of
[9:14] adoption of something that's coming out
[9:17] is so fast, right? like where um the
[9:20] time just gets compressed for the
[9:23] ingestion period for humanity on how
[9:25] quickly you actually learn something.
[9:26] It's because machines are now learning
[9:29] the language of humans rather than the
[9:31] other way around. And so everything gets
[9:33] exponentially faster as a result of it.
[9:36] Yeah, it's it's incredible how fast it
[9:38] is all moving. In part because of that
[9:40] speed though, many of us are concerned
[9:43] about what AI will mean for our work and
[9:47] for our career prospects and that of our
[9:49] children, right, and their children. The
[9:52] CEO of Anthropic has warned that AI
[9:54] could wipe out half of all entrylevel
[9:57] white collar jobs. He said it could
[9:59] increase unemployment to 10 to 20% and
[10:02] he said that could happen in the next 1
[10:03] to five years.
[10:05] What do you say to those who are
[10:07] concerned about AI wiping out a
[10:11] significant number of jobs?
[10:14] So, you know, Dario is is a friend. U we
[10:17] are investors in anthropic. Um I have a
[10:19] ton of respect for what he's done. I
[10:21] think in this area though, I have a
[10:23] slightly different opinion on a couple
[10:25] different dimensions. one is um
[10:31] the job and the configuration will
[10:34] change but like I was saying earlier I
[10:36] just refuse to believe that humans are
[10:38] going to be obsolete like it just seems
[10:40] like it's it's kind of an absurd concept
[10:42] uh we are far too creative for us to be
[10:45] obsolete you know um the second thing is
[10:48] this notion of entrylevel jobs going
[10:50] away I actually have a huge concern with
[10:52] this kind of line of thinking
[10:55] take a look at someone who's 20 years
[10:57] old
[10:59] and see how they use AI and then observe
[11:03] how someone who's 28 years old uses AI.
[11:06] Not even like a 50-year-old like me,
[11:08] like a 28-year-old, right? And you'll
[11:10] see a stark difference. And what's the
[11:12] difference? The 28-year-old might
[11:15] actually use AI in a very taskoriented
[11:18] transactional way. Give me an answer for
[11:20] something that I'm asking you right now.
[11:22] Over time, you might automate workflows.
[11:25] The 20-year-old uses it as a sidekick
[11:27] and a companion a as a brainstorming
[11:29] partner. And when you actually start to
[11:34] have generations that are coming in that
[11:37] have a very different starting point of
[11:39] when they started using this technology.
[11:42] Um, you know, I had a I have a
[11:43] 14-year-old daughter. Like I said, she
[11:45] started using the iPad, shame on us, way
[11:49] too early in life. And um, she was like
[11:52] two years old or three years old. and
[11:54] she she just instinctively knew how to
[11:56] kind of vertically scroll, right? First
[11:59] time she saw a newspaper, all she did
[12:01] was keep tapping
[12:05] on the picture and couldn't understand
[12:06] why the picture wasn't zooming in, you
[12:09] know, because she had a different kind
[12:10] of starting point orientation to this
[12:13] whole thing. And that's what you want to
[12:14] make sure that you inject into the
[12:17] workforce. I think aging and death, as
[12:20] glib as it sounds, is actually a really
[12:22] good thing because it makes way for the
[12:25] new to come in, right? And so if you
[12:27] just say, "I'm going to eradicate all
[12:29] entrylevel jobs," that's the stupidest
[12:32] thing a company can do in the long term
[12:34] because what you've done is you've
[12:36] actually taken away the injection of new
[12:40] perspective. And I think sometimes
[12:42] experience can be a massive liability. I
[12:45] think in there certain jobs where
[12:46] experience is really really valuable in
[12:48] certain jobs where experience can
[12:50] actually hold you back because how many
[12:52] times have all of us said at some point
[12:53] in time you know what we tried this out
[12:55] five years ago it didn't work so we're
[12:56] not going to do it again like 50,000
[12:59] assumptions might have changed from 5
[13:00] years ago when you tried it out but your
[13:03] experience in that point instinctively
[13:05] held you back so what you have to do is
[13:07] you have to try to figure out mechanisms
[13:09] for unlearning what you've learned over
[13:12] time and the the the early in career
[13:16] generation actually allows us to
[13:18] unlearn. So I spend an enormous amount
[13:21] of my time with student interns and
[13:23] people who are actually I I will put it
[13:26] under the mentorship guys because
[13:28] everyone will say okay I want to get
[13:29] mentored by someone who's you know
[13:32] relatively reasonably successful but I I
[13:35] get mentored as much by them as they get
[13:37] mentored by me because I learn a lot
[13:39] from my 14-year-old daughter. I learn a
[13:41] lot from people who've just gotten out
[13:42] of college because they have a fresh and
[13:45] a unique perspective and that
[13:46] perspective when coupled with experience
[13:50] makes magic happen. So I feel like it
[13:53] would be a really bad strategy to not
[13:55] have early and career people and entry-
[13:57] level people injected in your workplace.
[14:00] I don't think you should do that. I also
[14:02] don't think AI is only about
[14:04] productivity. In fact, that's in my mind
[14:06] the least interesting things about AI is
[14:09] getting productive. I think the more
[14:11] interesting things about AI is the
[14:12] quality of problems that you can solve
[14:15] that you couldn't even imagine solving
[14:16] before because that perspective didn't
[14:19] exist in the human corpus of knowledge
[14:21] which now does or now could and as a
[14:24] result you'll be able to do a whole lot
[14:26] more and I think a big part of that is
[14:28] the injection of new perspective that is
[14:31] not tainted with experience
[14:34] that make sense does I I I love the
[14:36] newspaper uh example that you uh that
[14:38] you gave. Um, so I know you said that
[14:41] you don't think that AI is going to make
[14:43] humans obsolete and that it should
[14:45] >> I'm curious how many over here think
[14:47] that you should just eradicate entry-
[14:48] level jobs. Anyone?
[14:51] Great. So, you think it's an absurd
[14:52] concept, right? Okay, great. I hope it's
[14:55] a yes for that one. How many think it's
[14:57] an absurd concept? The show of hands.
[15:00] Thank you. Okay.
[15:02] >> But you did hear Fay say during the last
[15:05] session
[15:06] >> that there could be there probably will
[15:08] be some growing pains like there is
[15:10] anytime there's a major new technology
[15:13] that's being adopted.
[15:15] >> What do those growing pains look like in
[15:17] in in your in your mind because I I
[15:19] understand what you're saying about
[15:19] entry level work and not making humans
[15:21] obsolete but it it does feel like
[15:23] there's going to be some significant
[15:25] impact
[15:25] >> there. there's meaningful impact. We as
[15:27] a society have to collectively make sure
[15:29] that the transition is a smooth one. Um,
[15:32] and by the way, every disruption that's
[15:33] happened, you've actually seen growing
[15:35] pains where people will actually get
[15:36] disrupted. You know, PE in tech, we live
[15:38] in a bubble. So, we keep thinking, oh,
[15:40] in tech disruption is just part of it.
[15:42] But when a steel mill worker gets
[15:44] disrupted, they don't become an AI
[15:47] prompt engineer,
[15:50] you know. Um, and so what you start to
[15:52] see is there's there's actually a level
[15:56] of retraining and reskilling in society
[15:59] that has to be done that the tech
[16:02] community has to actually take some
[16:03] responsibility for and we have to work
[16:06] closely with the with the public sector
[16:08] and make sure that education and the
[16:10] skills gap is actually narrowed because
[16:12] if you don't do that you will you will
[16:15] create um you know some level of pain in
[16:17] society and we we want we want to make
[16:18] sure we avoid that. The second thing
[16:20] that I think is pretty important is
[16:22] safety and security are going to be
[16:25] pretty massive consequential and
[16:27] potentially catastrophic outcomes if not
[16:31] um if not mitigated effectively. For
[16:33] example,
[16:35] um every tool that we use for AI is the
[16:38] same tool that an adversary is going to
[16:40] be able to use. That adversary is going
[16:42] to have attacks that are going to be far
[16:45] more sophisticated. Those adversaries
[16:47] might be nation states. uh those attacks
[16:50] might be far more bespoke where you used
[16:52] to have, you know, an exotic prince from
[16:54] an exotic country that was asking you to
[16:57] give you their bank account so they
[16:58] could just transfer $10 million with a
[17:00] bunch of typos in an email which was a
[17:02] spam email.
[17:04] >> Now might be someone that says, "Hey, it
[17:06] was great to see you at the our kids
[17:08] basketball game last night. Here are
[17:09] some pictures. Click to download." And
[17:11] the moment you clicked, there's a piece
[17:12] of malware that gets downloaded on your
[17:14] machine that can then actually start
[17:16] having lateral movement. So what you
[17:18] need to do is make sure that safety and
[17:20] security is something that is pretty
[17:22] importantly dealt with and that has to
[17:24] be dealt with not just by using AI for
[17:27] cyber defense. You have to secure AI
[17:30] itself because these models they
[17:32] hallucinate. These models they have
[17:33] toxicity. These models can have prompt
[17:36] injection attacks. And so how do you
[17:38] make sure that those things can get
[17:39] mitigated? And so the way in which you
[17:43] get visibility on what data is going
[17:45] into a model, the way in which you
[17:47] actually validate a model so that the
[17:48] model is working the way that you expect
[17:50] it to work and the way you put runtime
[17:53] enforcement guard rails, the way that an
[17:55] agent has a different kind of identity
[17:58] that you can validate that can talk to
[18:00] another agent but has fine grain control
[18:02] because you don't want to have an agent
[18:04] say, "Okay, get access to my email and
[18:06] send an email to everyone." You might
[18:08] say, "I want to have an agent be able to
[18:10] craft an email on my behalf, but not to
[18:12] my boss and my board of directors, but
[18:14] it's okay to do it to my team." And
[18:16] those are the kind of things that we
[18:17] have to make sure that we actually think
[18:19] through. And that risk mitigation on
[18:21] safety and security will be a pretty big
[18:22] one that we as a collective society have
[18:24] to solve for as well.
[18:25] >> Yeah, absolutely. Thank you so much, Ju,
[18:28] for doing this. Uh we really appreciate
[18:29] all the insights. This was great.
[18:31] >> Thank you so much. Cheers.
