The most hyped AI breakthrough of 2026 is a hack

Summary
In this video, presenter Brendan Dell critically analyzes OpenAI’s September 2026 claim of solving the Navier–Stokes Millennium Prize problem. He examines the backstory behind the proof—including prior human mathematical breakthroughs, a priority dispute involving NYU mathematician Tristan Buckmaster and Anthropic researcher Levent Alpoge, and the PR packaging surrounding the announcement. Dell argues that while AI systems excels at grinding through formally checkable verification tasks, they lack human mathematical understanding and scientific judgment.

What is shown

• [00:01] News coverage and commentary displayed from the BBC, Scientific American, NPR, and Numberphile.
• [00:27] Screenshots of OpenAI’s blog post detailing the multi-agent solving process, agent message counts, and pass-rate graphs comparing GPT-6 Astra against an internal model.
• [01:40] Comparison of OpenAI’s viral, stylized vortex render against Matt Parker’s (Stand-up Maths) breakdown and the actual schematic diagram in OpenAI’s 165-page preprint (Finite Time Blowup for Navier–Stokes).
• [03:01] Clips from Brady Haran’s Numberphile interview with mathematician Tristan Buckmaster discussing convex integration and his private communications with OpenAI researcher Sébastien Bubeck.
• [04:03] Overview of the Clay Mathematics Institute’s Millennium Prize Problems website and rules.
• [07:03] The open statement A Severe Misalignment of AI in Mathematics, signed by 28 Fields Medalists.
• [08:47] Historical case study of Alan Turing and David Hilbert’s Entscheidungsproblem via the Stanford Encyclopedia of Philosophy.
• [10:42] Promotional segment for Dell’s training course, The Sovereign Professional.
• [13:17] Interface and documentation for the Lean interactive theorem prover.
• [14:26] Excerpt from Terence Tao’s blog detailing foundational work by Diego Córdoba and Luis Martínez-Zoroa.
• [17:16] Excerpts from the survey paper Recursive Self-Improvement in AI: From Bounded Self-Refinement to Autonomous Research Loops (Chen et al., July 2026) illustrating the “verification hierarchy.”

Claims & numbers

• The presenter notes that OpenAI claimed its system solved the Navier–Stokes existence and smoothness problem using approximately 10,000 concurrent agents in 88 hours, generating 2.7 million messages and roughly 130 billion output tokens (out of 4.9 million messages and 300 billion tokens across all attempted problems).
• The Clay Mathematics Institute established a $1,000,000 prize in 2000 for each of the seven Millennium Prize problems.
• According to Buckmaster’s account, Tristan Buckmaster and Levent Alpoge established a blowup result for the Euler and Boussinesq equations on August 15, 2026, formalizing and verifying it in Lean by August 22, 2026.
• OpenAI launched its multi-agent attempt on September 1, 2026, after rumors circulated in the tech industry that Anthropic had resolved the problem.
• OpenAI published its 165-page paper on September 8, 2026, following a call between Sébastien Bubeck and Buckmaster on September 6.
• On September 11, 2026, 28 Fields Medalists signed a declaration arguing that AI labs’ push to race through famous math problems as benchmarks is detrimental to the mathematical community.
• The presenter estimates that OpenAI spent upwards of $20 million in compute running its agent search swarm.

Notable quotes

• [04:46] “Bubeck had said that the machine did it with ‘very little human input.’ However, on the call, that turned out to not be true.”
• [09:47] “‘Technically, all the plot lines are resolved, but most of the value of the experience was lost,’ he said.” (quoting Terence Tao via IBM Think)
• [16:33] “The LLM does not understand the answer it gave us. And it is not because the LLM is so much smarter than humans; it’s because it can grind on a checkable process for as long as we will give it the resources to do so...”

Assessment
This is an independent critical commentary and review video examining the academic and PR controversy surrounding OpenAI’s Navier–Stokes proof. The creator uses published papers, blog posts, and third-party interviews (Numberphile, Stand-up Maths) to substantiate his critique, while incorporating a mid-video sponsorship pitch for his own career consultancy program.

Described by gemini-3.8-flash on 2026-10-08 from the video’s audio and frames.