OpenAI Claims Navier-Stokes Proof, Sparking Pure-Math Backlash

Sabine Hossenfelder

OpenAI's 160-page Navier-Stokes proof likely holds up despite circular-reasoning risks, but private labs will abandon pure mathematics because the breakthroughs generate no commercial revenue.

AI-assisted summary of AI Solved a Millennium Problem and Everyone's Angry — original source. Check the original for full context.

Source published: 2026-09-09

Key ideas

OpenAI has published a 160-page proof claiming to solve the Navier-Stokes Clay Millennium Prize Problem by demonstrating a finite-time blow-up using an engineered adaptive forcing term. (0:00)
The academic community has responded with widespread frustration, driven largely by resentment that a private corporate AI lab solved a historic mathematical milestone rather than human academics. (0:00)
While arithmetic errors were mitigated by cross-verifying AI agents, circular reasoning remains the primary mathematical vulnerability in large language model proofs. (1:43)
Sabine Hossenfelder estimates a 90% probability that OpenAI's proof is mathematically valid, but expects corporate AI labs to pivot away from pure math due to a lack of commercial monetization. (4:28)

Detailed notes

## Background Published on YouTube on September 9, 2026, by physicist and creator Sabine Hossenfelder, this video shares her thoughts on OpenAI's reported solution to the Navier-Stokes Clay Millennium Prize Problem. The video addresses both the technical details of the mathematical claim and the heated academic backlash triggered by a private tech lab solving a celebrated prize problem.

## TL;DR Sabine Hossenfelder examines OpenAI's 160-page proof demonstrating a finite-time singularity in the Navier-Stokes equations via an inverse formulation with engineered, adaptive forcing pulses. She addresses the bitter resentment among human academics who feel their territory was poached, while also warning about subtle risks of circular reasoning inherent in frontier large language models. Defending the problem's abstract value through its ties to black hole physics and general relativity, she ultimately predicts the proof is likely sound, but notes that private AI labs will soon abandon pure mathematics due to a total lack of commercial payoff.

## Atmosphere & style The presentation is dry, sardonic, and unapologetically direct, capturing Hossenfelder's signature deadpan commentary on academic culture. It carries an amused weariness toward academic gatekeeping, contrasting the wounded pride of human researchers against the indifferent momentum of corporate AI deployment. Listening to it feels like an insider pull-back of the curtain on scientific vanity, pairing accessible mathematical intuition with blunt, unromantic realism about corporate incentives.

## Bulletin - [00:00](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=0s) OpenAI has published a 160-page proof claiming to solve the Navier-Stokes Clay Millennium Prize Problem by demonstrating a finite-time blow-up using an engineered adaptive forcing term. - [00:00](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=0s) The academic community has responded with widespread frustration, driven largely by resentment that a private corporate AI lab solved a historic mathematical milestone rather than human academics. - [01:43](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=103s) While arithmetic errors were mitigated by cross-verifying AI agents, circular reasoning remains the primary mathematical vulnerability in large language model proofs. - [04:28](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=268s) Sabine Hossenfelder estimates a 90% probability that OpenAI's proof is mathematically valid, but expects corporate AI labs to pivot away from pure math due to a lack of commercial monetization.

## Key takeaways - [00:00](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=0s) OpenAI announced a claimed solution to the Navier-Stokes Millennium Prize Problem, sparking strong negative pushback from academic mathematicians and physicists. - [00:00](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=0s) The core question of the Navier-Stokes problem is whether solutions remain smooth indefinitely or develop finite-time singularities (blow-ups). - [00:00](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=0s) Valid solutions with external forcing require proving that the forcing function remains strictly bounded and regular so infinite energy is not pumped into the system. - [01:43](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=103s) OpenAI reverse-engineered their solution by positing a fluid field with an inherent singularity, inserting it into the equation, and defining the mathematical remainder as the forcing term. - [01:43](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=103s) The proof's physical mechanism relies on an engineered vortex continuously stretched and focused along its central axis via state-adaptive external pulses. - [01:43](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=103s) Most of the 160-page paper consists of technical bounds and lemmas proving that the adaptive driving pulses contribute finite energy overall. - [01:43](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=103s) The greatest technical risk in AI-generated mathematical proofs is circular reasoning, where the model inadvertently assumes the bounds it is supposed to establish. - [01:43](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=103s) The proof's structural approach is plausible because stretched axial vortices have long been considered primary blow-up candidates by human mathematicians. - [04:28](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=268s) Navier-Stokes fluid dynamics connects fundamentally to general relativity, particularly the fluid dualities governing black hole event horizons. - [04:28](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=268s) Pure mathematical breakthroughs yield little direct commercial revenue beyond prize money, which will likely limit sustained corporate interest in pure math. - [04:28](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=268s) OpenAI's reported interest in room-temperature superconductors represents a significantly harder challenge for AI due to experimental messiness and a lack of clean axiomatic frameworks. - [04:28](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=268s) Sabine assesses a 90% chance the proof stands up to mathematical scrutiny alongside a near-certainty that academic complaints will persist.

## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=0s) The Navier-Stokes Millennium Problem and the Academic Backlash

OpenAI has publicly claimed to have solved the Navier-Stokes Millennium Prize Problem. Having worked on the very same problem over the past year using AI models—as many others undoubtedly have—Sabine Hossenfelder expresses surprise at how overwhelmingly negative the academic reaction has been. Rather than showing genuine curiosity about the mathematical machinery, how OpenAI engineered the solution, or what new avenues this opens up, mathematicians and physicists seem primarily upset that a private AI lab poached a milestone that "humans" were supposed to resolve. For decades, the academic community complained that private corporations showed zero interest in pure mathematics; now that a private company has taken an active interest and produced a result, academics are furious that they didn't leave it alone.

To unpack what is at stake, the core mathematical challenge asks whether a realistic fluid governed by the Navier-Stokes equations can develop singularities—or "blow-ups"—in finite time. The Navier-Stokes system serves as the foundational mathematical framework describing realistic fluid flows and gases, underpinning critical real-world applications such as modern weather forecasting. In the standard formulation:

$$\frac{\partial \mathbf{u}}{\partial t} + (\mathbf{u} \cdot \nabla)\mathbf{u} = -\nabla p + \nu \nabla^2 \mathbf{u} + \mathbf{f}$$

where $\mathbf{u}$ represents the fluid velocity field, $p$ is the internal pressure, $\nu$ (nu) denotes the kinematic viscosity constant, and $\mathbf{f}$ represents the external forcing applied across space and time. Because a fluid can trivially blow up if subjected to an arbitrarily massive external force, any valid solution requires proving that the external forcing itself stays strictly regular and bounded. The purest and notoriously most difficult route is setting the forcing strictly to zero ($\mathbf{f} = 0$), whereas OpenAI tackled the problem through an alternative formulation.

- **The Problem Core**: Demonstrating whether solutions to the 3D Navier-Stokes equations remain smooth for all time or develop finite-time singularities (blow-ups). - **The Forcing Condition**: While zero forcing ($\mathbf{f}=0$) is the most rigorous formulation, solving it with an engineered forcing requires proving that the forcing function remains strictly finite and regular.

---

[01:43](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=103s) Inside OpenAI's 160-Page Proof: Engineered Pulses and Circular Traps

OpenAI approached the problem inversely: rather than evolving an unforced fluid toward an organic blow-up, they began by postulating a fluid field configuration that inherently features a singularity. Plugging this target field directly into the Navier-Stokes equation, they labeled the resulting mathematical remainder as the "forcing" term ($\mathbf{f}$). Under the official rules of the Clay Millennium Prize, this approach is entirely legitimate, provided one can rigorously prove that the required forcing does not pump infinite energy into the fluid system. The resulting proof spans roughly 160 pages of dense estimates and bounds.

```mermaid flowchart TD accTitle: OpenAI Navier-Stokes Proof Architecture accDescr: Flowchart illustrating the reverse construction method of OpenAI's Navier-Stokes blow-up proof.

subgraph Target["Target Construction"] A["Construct Fluid Field u with Singularity"] --> B["Insert into Navier-Stokes Equation"] B --> C["Define Remainder as Forcing f"] end

subgraph Dynamics["Engineered Driving"] C --> D["Inject Pulses at Adaptive Time Intervals"] D --> E["Vortex Stretches & Focuses on Center Axis"] end

subgraph Verification["Mathematical Burden (160 Pages)"] E --> F["Estimate Total Contribution of Forcing"] F --> G["Lemma-by-Lemma Proof: Finite Energy & Regularity"] end ```

The underlying physical model relies on supplying the fluid with precisely timed external pulses that adapt dynamically to the fluid’s instantaneous state, actively driving it toward a catastrophe. The specific singularity analyzed is an engineered vortex that continuously stretches and concentrates along its central axis. The bulk of the 160-page paper is an exhausting chain of lemmas bounding the total energy contribution of these pulses and managing cumulative error bounds.

While OpenAI deployed thousands of AI agents to cross-verify the mathematical steps, making trivial arithmetic or step-by-step errors unlikely, a far subtler danger persists: circular reasoning. In extensive hands-on experience with models such as GPT-4, GPT-5-era iterations, and Google's Astra, large language models frequently fall into traps where they inadvertently assume the very regularity condition or bound they are tasked with establishing:

> "It's like saying if pigs can't fly, then the laws of gravity are broken. And if the laws of gravity are broken, then pigs can fly. And that's totally correct, but it doesn't prove that either pigs fly or the laws of gravity were broken, just that both are consistent with each other."

- **Proof Architecture**: Approximately 160 pages dedicated to verifying that the adaptive forcing pulses do not inject unlimited energy. - **Model Comparison**: Even advanced iterations like GPT-5.6 struggle to avoid subtle circular assumptions in deep mathematics, while models like Astra appear noticeably weaker at high-level mathematical reasoning. - **Plausibility**: Finite-time blow-ups via stretched axial vortices represent the exact class of candidates mathematicians have explored over the last decade, making the overall claim structurally plausible even before full human verification.

---

[04:28](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=268s) Real-World Utility, Black Holes, and the Future of AI Science

Addressing dismissive claims that the proof is "useless" or that "no one cares," Sabine argues that utility in theoretical physics is often misunderstood. Outside the lure of the Clay Institute's $1,000,000 prize, the mathematical behavior of the Navier-Stokes equations shares profound dualities with general relativity, particularly the fluid dynamics governing black hole horizons under the membrane paradigm. To those who dismiss such pure curiosity, she counters: "Well, what's the use of kicking a ball into a net? Personally, I think if some people find it interesting, that's the use in itself."

Commercially, pure math proofs generate virtually zero revenue, which will likely cool OpenAI's long-term corporate appetite for pure mathematics. Reports suggest Sam Altman has already directed focus toward room-temperature superconductors—a drastically messier domain because it lacks the clean axiomatic rigor of pure mathematics and cannot be solved purely inside a formal symbolic sandbox.

Finally, addressing accusations that OpenAI may have derived its proof by unfairly leveraging or "sneaking in" on pre-existing academic preprints, Sabine dismisses academic gossip in favor of the larger existential dilemma facing research:

> "Do we need to keep people motivated by giving them a million-dollar prize, or do we just leave it to AI to begin with? And will AI companies continue to be motivated if they notice there's no money to make with it?"

She concludes with an asymmetric forecast: there is a 90% chance the proof holds up mathematically, but a 99% chance that human mathematicians will never stop complaining about it.

- **Cross-Disciplinary Importance**: Navier-Stokes regularity provides mathematical insights into general relativity and black hole event horizon physics. - **Commercial Reality**: Aside from a one-time cash prize, pure mathematical breakthroughs cannot be monetized, making a sustained corporate AI push into pure mathematics unlikely. - **Next Frontier**: The pivot toward room-temperature superconductivity presents a fundamentally harder problem for AI agents due to experimental noise and undefined mathematical foundations.

## Notable quotes > [01:43](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=103s) It's like saying if pigs can't fly, then the laws of gravity are broken. And if the laws of gravity are broken, then pigs can fly. And that's totally correct, but it doesn't prove that either pigs fly or the laws of gravity were broken, just that both are consistent with each other. — Sabine Hossenfelder

> [04:28](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=268s) Well, what's the use of kicking a ball into a net? Personally, I think if some people find it interesting, that's the use in itself. — Sabine Hossenfelder

> [04:28](https://www.youtube.com/watch?v=eR2q8s4VB1w&t=268s) Do we need to keep people motivated by giving them a million-dollar prize, or do we just leave it to AI to begin with? And will AI companies continue to be motivated if they notice there's no money to make with it? — Sabine Hossenfelder

## Mentioned OpenAI, Sabine Hossenfelder, Navier-Stokes Millennium Prize Problem, Clay Mathematics Institute, GPT-4, GPT-5, GPT-5.6, Astra, Google, Sam Altman

## Video description Some thoughts on OpenAI's proof of the Navier Stokes Clay Millenium Problem

👕T-shirts, mugs, posters and more: ➜ https://sabines-store.dashery.com/ 💌 Support me on Donorbox ➜ https://donorbox.org/swtg 👉 Transcript with links to references on Patreon ➜ https://www.patreon.com/Sabine 📝 Transcripts and written news on Substack ➜ https://sciencewtg.substack.com/ 📩 Free weekly science newsletter ➜ https://sabinehossenfelder.com/newsletter/ 👂 Audio only podcast ➜ https://open.spotify.com/show/0MkNfXlKnMPEUMEeKQYmYC 🔗 Join this channel to get access to perks ➜ https://www.youtube.com/channel/UC1yNl2E66ZzKApQdRuTQ4tw/join 📚 Buy my book ➜ https://amzn.to/3HSAWJW

[Source](https://www.youtube.com/watch?v=eR2q8s4VB1w)

中文简报

OpenAI宣称证明纳维-斯托克斯方程，引发纯数学界强烈反弹 OpenAI长达160页的纳维-斯托克斯证明尽管存在循环论证风险，但可能成立；然而由于此类突破无法产生商业收益，私立实验室或将放弃纯数学研究。 Sabine Hossenfelder审视了OpenAI长达160页的证明，该证明通过带有工程化自适应强迫脉冲的逆向公式，展示了纳维-斯托克斯方程在有限时间内的奇点。她探讨了人类学者因领地被侵占而产生的强烈不满，同时也对前沿大语言模型固有的微妙循环论证风险提出了警告。在通过与黑洞物理学和广义相对论的联系来捍卫该问题的抽象价值的同时，她最终预测该证明很可能成立，但指出由于完全缺乏商业回报，私立AI实验室很快将放弃纯数学领域。 OpenAI发表了一篇长达160页的证明，宣称通过使用工程化自适应强迫项展示有限时间内的爆破，解决了纳维-斯托克斯克雷千禧年大奖难题。 (0:00) 学术界反应普遍沮丧，这主要是因为对一家私立商业AI实验室而非人类学者攻克了这一历史性数学里程碑感到愤懑不平。 (0:00) 尽管算术错误已通过交叉验证的AI智能体得以减少，但循环论证仍然是大语言模型证明中最主要的数学漏洞。 (1:43) Sabine Hossenfelder估计OpenAI的证明有90%的概率在数学上有效，但预计商业AI实验室将因缺乏商业变现途径而转向离开纯数学领域。 (4:28)

Reading summary

Sabine Hossenfelder evaluates OpenAI's claim of solving the Navier-Stokes Millennium Prize Problem, arguing that while the proof's reverse-engineering method is structurally plausible, it highlights an uncomfortable shift as private AI labs enter non-commercial pure mathematics. Academic Discontent Over Corporate Mathematics (0:00) The 3D Navier-Stokes equations describe the motion of fluid flows and gases, forming the mathematical foundation for fields like weather forecasting. The Millennium Prize Problem asks whether solutions to these equations remain smooth indefinitely or develop finite-time singularities (blow-ups). While the purest mathematical formulation sets external forcing to zero ($f = 0$), the official rules of the Clay Millennium Prize permit solutions that include an external forcing term, provided that the forcing remains strictly finite and regular without injecting infinite energy into the fluid system. OpenAI's Reverse Proof and Circular Reasoning Risks (1:43) OpenAI approached the problem through an inverse construction rather than tracking an unforced fluid toward an organic blow-up. They postulated a target fluid configuration that inherently contains a singularity, inserted it into the Navier-Stokes equations, and treated the remaining mathematical terms as an engineered forcing function. This forcing relies on adaptive pulses that continually stretch and concentrate an axial vortex until it blows up. The accompanying 160-page proof consists primarily of lemmas establishing cumulative error bounds and proving that the adaptive driving pulses inject only finite energy. While OpenAI used thousands of AI agents to check mechanical steps, Hossenfelder notes that large language models—including iterations of GPT and Google's Astra—remain prone to circular reasoning. In high-level mathematics, these models risk inadvertently assuming the regularity conditions or bounds they are tasked with establishing, creating proofs where premises and conclusions are internally consistent without genuinely proving the target theorem. Theoretical Significance and Commercial Limits (4:28) Addressing dismissals that the proof lacks practical application, Hossenfelder emphasizes that Navier-Stokes mathematics shares theoretical dualities with general relativity, specifically the fluid dynamics describing black hole horizons under the membrane paradigm. However, she notes that pure mathematical proofs offer virtually no commercial revenue beyond the $1,000,000 Clay Institute prize. Consequently, corporate AI research may pivot away from pure mathematics toward messier, high-value domains like room-temperature superconductors, which cannot be solved within an idealized symbolic sandbox. Hossenfelder assigns a 90% probability that OpenAI's mathematical proof will withstand formal verification, alongside a 99% probability that academic complaints will persist. The dispute raises unresolved questions about whether monetary prizes will still motivate human researchers and whether AI labs will sustain interest in non-monetizable fundamental science.

Source quotes

[1:43] Sabine Hossenfelder: It's like saying if pigs can't fly, then the laws of gravity are broken. And if the laws of gravity are broken, then pigs can fly. And that's totally correct, but it doesn't prove that either pigs fly or the laws of gravity were broken, just that both are consistent with each other. [4:28] Sabine Hossenfelder: Well, what's the use of kicking a ball into a net? Personally, I think if some people find it interesting, that's the use in itself. [4:28] Sabine Hossenfelder: Do we need to keep people motivated by giving them a million-dollar prize, or do we just leave it to AI to begin with? And will AI companies continue to be motivated if they notice there's no money to make with it?

Original transcript

[0:00] OpenAI claims to have solved the [0:02] Navier-Stokes Millennium Problem. It's [0:04] not a secret that I've been working on [0:06] the same problem in the past year [0:08] without my eyes models, and I'm sure so [0:10] have many others. So, what are we to [0:12] make of the solution? And is it really [0:14] as useless as the critics complain? Let [0:17] me say how that I've been struck by how [0:20] extremely negative the reaction has been [0:23] to the OpenAI proof. Like, you'd think [0:25] that mathematicians and physicists would [0:27] be interested in how the proof works, [0:30] how OpenAI managed to get it done, what [0:33] else can be done with it. Instead, [0:36] everyone's complaining that OpenAI [0:37] didn't leave it to humans. For decades, [0:40] they've complained that private business [0:42] has no interest in pure maths. Now, [0:44] they're complaining private business [0:46] does have an interest in pure maths. [0:48] Okay, but first, let me tell you what [0:50] we're even talking about. The problem to [0:52] be solved is the question of whether a [0:55] fluid, that's described by the [0:57] Navier-Stokes equation, can develop [1:00] singularities or blow-ups in finite [1:03] time. The Navier-Stokes equation is the [1:06] key equation that describes realistic [1:09] fluids and gases, and it's, for example, [1:11] used for the weather forecast. The [1:13] equation looks like this, where U is the [1:16] fluid itself, and P is the pressure. [1:18] This new is the viscosity, that's a [1:20] constant, and this F here is what's [1:23] called the forcing, that's the force [1:25] that acts on the fluid in each point. [1:28] Now, it's, of course, possible to get [1:29] the fluid to blow up if you make the [1:31] forcing large enough. So, if you want to [1:32] solve the problem, you have to show that [1:34] the forcing itself remains regular. The [1:37] cleanest way to do this is to use zero [1:40] forcing. That's also the hardest way. [1:43] The OpenAI proof now, instead, starts [1:45] with a fluid field that has a blow-up. [1:48] You just insert this into the equation [1:50] and call whatever's left the forcing. [1:53] Then, you have to prove that the [1:54] remaining forcing itself remains [1:56] regular. The entire proof is 160 pages [1:59] or so. I did not check it. I'll leave [2:01] that to competent people, but I can make [2:03] some incompetent observations. The [2:06] construction works by supplying the [2:09] fluid with pulses that come in at [2:11] specifically chosen time intervals that [2:14] adapt to the fluid dynamics. So, it's a [2:16] deliberately engineered driving of the [2:18] fluid into a singularity. The simple [2:21] singularity they looked at is a type of [2:23] vortex that becomes increasingly [2:25] stretched and focused on its center [2:27] axis. This is what the image shows. This [2:30] is totally allowed by the criteria of [2:32] the millennium problem, but it puts the [2:34] burden on having to prove that the [2:36] forcing does not supply unlimited [2:38] energy. This means they have to somehow [2:40] keep track of what the pulses do and sum [2:43] them all up and estimate the errors and [2:45] so on. This is what most of the proof's [2:48] about. It's lemma after lemma about how [2:51] to estimate the total contribution from [2:53] the forcing. Now, look going by what [2:56] OpenAI said, they had literally [2:58] thousands of AI agents working on this. [3:00] So, it seems highly unlikely to me that [3:02] the proof is directly wrong. However, [3:05] what I've noticed is much more likely to [3:07] happen with the publicly available [3:09] OpenAI models is that they run into [3:11] circular arguments where they implicitly [3:14] assume what needs to be proved. They [3:16] might, for example, assume a certain [3:18] regularity or solvability condition for [3:21] a lemma that actually contained the [3:23] bound they had to prove. The issue is [3:25] that this sort of argument has no actual [3:28] mistake in the step-by-step procedure. [3:30] You see, it's like saying if pigs can't [3:32] fly, then the laws of gravity are [3:34] broken. And if the laws of gravity are [3:35] broken, then pigs can fly. And that's [3:38] totally correct, but it doesn't prove [3:40] that either pigs fly or the laws of [3:42] gravity were broken, just that both are [3:44] consistent with each other. So, it might [3:47] be that the proof is consistent in that [3:50] sense and yet circular in not proving [3:52] the thing they need to prove. I found [3:55] that even GPT-5.6 still struggles with [3:58] this. By the way, my impression is that [4:00] the new Astra is worse than GPT-5.6 when [4:03] it comes to maths, but that's just an [4:06] aside. Let me put it like this. [4:08] Personally, I think that the [4:09] Navier-Stokes equation does have [4:11] blow-ups, and the class that they looked [4:13] at in the proof is the case that most [4:16] efforts in the past decade have focused [4:18] on. So, it's plausibly correct. Still, [4:21] I'll not take it for granted that the [4:24] proof's correct until humans have [4:26] checked it. Then, let me say something [4:28] about the question of whether this is [4:31] relevant. I see a lot of people [4:33] commenting, "Oh, no one cares. This [4:35] isn't useful for anything." and so on. [4:38] Well, the reason I've been interested in [4:40] this, besides the million-dollar prize, [4:42] is that the Navier-Stokes equation has a [4:44] known relation to some problems in [4:46] general relativity, especially [4:49] concerning the physics of black hole [4:51] horizons. Now, you could say again, [4:53] "Okay, but this isn't useful for [4:55] anything." at which point I'd like to [4:57] ask, "Well, what's the use of kicking a [4:59] ball into a net?" Personally, I think if [5:02] some people find it interesting, that's [5:04] the use in itself. Can you make money [5:07] with it? I mean, other than winning this [5:09] prize? Probably not. And I guess that [5:12] this will soon dampen Oop May I ask [5:14] maths enthusiasm. But some Altmans are [5:16] here now be looking at room temperature [5:18] superconductors, which I think will keep [5:20] GPT occupied for a while. That problem's [5:24] on a completely different level because [5:26] it's not mathematically well-defined to [5:29] begin with. Finally, let me add that I'm [5:32] aware there's some controversy about [5:34] what I wrote May I ask model sneaked in [5:36] on someone else's research. I have to [5:39] admit that I'm super not interested in [5:41] people stories. I really just care about [5:44] science. You might say, "Okay, but if [5:47] you care about science, you need to keep [5:49] people motivated to do it, so you must [5:51] care about people. And if their results [5:54] get stolen, they won't continue doing [5:56] it." Well, I guess that's the question [5:59] we have to answer. Do we need to keep [6:02] people motivated by giving them a [6:04] million-dollar prize, or do we just [6:07] leave it to AI to begin with? And will [6:10] AI companies continue to be motivated if [6:13] they notice there's no money to make [6:15] with it. In summary, I think there's a [6:17] 90% chance this proof is correct, but a [6:20] 99% chance that mathematicians will [6:22] continue to complain about it. If you [6:25] haven't yet reached your daily limit of [6:27] uninformed opinions about useless [6:28] topics, please subscribe and buy a [6:31] t-shirt. [6:32] Thanks for watching.