Will AI systems that solve advanced formal problems like frontier mathematics soon drive transformative practical advances across most scientific and engineering fields?
The dispute is whether AI systems that can solve hard math and other formal problems will soon drive transformative advances across most sciences and engineering in discussions since early September. Wissner-Gross and Ismail on Moonshots claim AI has effectively solved mathematics and will bulk-solve many other fields, rapidly boosting medicine and engineering. Millidge on Dwarkesh Podcast and Sacks on All-In argue these feats mainly showcase reasoning or computational power and will not soon yield breakthroughs like new medicines or aircraft.
Such AI capabilities will soon transform progress in science, engineering, medicine, and other fields.
Wissner-Gross (Moonshots) claims AI is scalably solving the entire field of mathematics and will reveal many previously unknown results across the mathematical universe. Blundin (Moonshots) argues that once AI has solved mathematics, engineering design problems become fast lookups of stored answers, so engineering progress accelerates sharply. Wissner-Gross states that after mathematics, AI will push an explosive wavefront of bulk problem-solving through medicine, with many medical questions solved in the same way. Ismail (Moonshots) claims that in any field where all possible problems can be finitely enumerated, AI will repeatedly build databases of precomputed solutions.
6 claims, in their words
-
“I think we're going to see this play out over and over again for every single field where every possible problem can be enumerated in a finite number of problems.”
-
“AI has solved the field. AI is scalably solving the field. And this is exactly what we want. The, the field of math is set to flourish and be flourished because we're about to learn the answers to the mysteries of the mathematical universe”
-
“the explosive wavefront of solving everything at medicine next. ... It's just going to play out the same way.”
-
“Mathematicians as artisans. It's over. It's cooked. It's been flourished, if you prefer. It's over. Math is over, as we've, as it has been done historically. The new way of doing math is bulk solution.”
-
“we're going to start to see this play out over and over again.”
-
“Once math is all solved, it's a simple lookup of the answers. So the, the AIs are going to move at warp speed through all these engineering problems. ... the pace of engineering grows massively with this.”
These abilities show impressive reasoning but will not soon transform practical science and engineering.
Sacks (All-In) argues that recent AI-generated mathematical proofs will not directly produce breakthroughs like a cancer cure or a new supersonic airplane. He claims these proofs mainly demonstrate AI computational horsepower and recursive improvement, rather than immediate contributions to practical technology. Sacks reports that when he asked ChatGPT about these proofs, it predicted no direct near-term innovations to everyday life from them. Millidge (Dwarkesh Podcast) argues that past AI achievements in domains like math competitions or chess often led to less real-world impact than observers expected, echoing Moravec-style surprises.
3 claims, in their words
-
“there's been a classic thing, almost like Marvex paradox, right? Where, like, we see, like, you know, we think of the AI being like, if it can do this, it's going to be amazing, right? Like, if it can solve these hard math problems, if it can win a chess, blah, blah. And then it solves these things, and then”
-
“I think that this, I think this class of proof is, it's really great. It shows computational horsepower overcame other things, as you said, Friedberg, because you can recursively improve to get there. But I don't think these proofs is what is going to necessarily show you that there's a cure to cancer, or there's a supersonic plane.”
-
“if it's worth saying it, I asked ChatGPT, and I said, will the latest proofs of important math problems result in direct Improvements and innovations to life as we know it, be honest and don't extrapolate weakly. And its conclusion was the latest mathematical proofs are more compelling as evidence that AI is acquiring sophisticated reasoning capabilities than as evidence”
Listeners of “These abilities show impressive reasoning but will not soon transform practical science and engineering.” have not heard
Wissner-Gross's claim that traditional artisanal mathematics is over and being replaced by AI bulk solution suggests a discontinuous change in how formal reasoning contributes to future technologies. His view that AI bulk-solving will repeat across entire domains, as it has in mathematics, presents a picture of cascading waves of problem resolution through many sciences. Blundin's claim that, after AI solves mathematics, many engineering problems reduce to rapid lookups of stored results implies a major speedup in designing complex systems.
3 unanswered arguments, in their words
-
“Mathematicians as artisans. It's over. It's cooked. It's been flourished, if you prefer. It's over. Math is over, as we've, as it has been done historically. The new way of doing math is bulk solution.”
-
“we're going to start to see this play out over and over again.”
-
“Once math is all solved, it's a simple lookup of the answers. So the, the AIs are going to move at warp speed through all these engineering problems. ... the pace of engineering grows massively with this.”
Listeners of “Such AI capabilities will soon transform progress in science, engineering, medicine, and other fields.” have not heard
Listeners of these shows have not heard Millidge's point that AI successes in math problems or chess have often produced less impact than expected, challenging assumptions about current math feats. Sacks's argument that recent AI-generated proofs show computational horsepower yet will not by themselves create technologies like a cancer cure or supersonic plane questions expectations of near-term practical transformation.
2 unanswered arguments, in their words
-
“I think that this, I think this class of proof is, it's really great. It shows computational horsepower overcame other things, as you said, Friedberg, because you can recursively improve to get there. But I don't think these proofs is what is going to necessarily show you that there's a cure to cancer, or there's a supersonic plane.”
-
“there's been a classic thing, almost like Marvex paradox, right? Where, like, we see, like, you know, we think of the AI being like, if it can do this, it's going to be amazing, right? Like, if it can solve these hard math problems, if it can win a chess, blah, blah. And then it solves these things, and then”
Wissner-Gross (Moonshots) and Ismail (Moonshots) say AI solved mathematics and will bulk-solve domains, accelerating engineering through lookups, points absent from Millidge's and Sacks's discussions. Sacks (All-In) and Millidge (Dwarkesh Podcast) say AI feats had limited payoff and current proofs lack direct paths to technologies, points not answered on Moonshots.
- Oct 20, 2026 Published: cleared the gate.
- Oct 20, 2026 David Sacks moved from “AIs are solving important open math problems like Millennium Prize problems.” to “These abilities show impressive reasoning but will not soon transform practical science and engineering.”.
- Oct 20, 2026 5 new claims (Dwarkesh Podcast, Moonshots).
- Oct 18, 2026 10 new claims (The Diary Of A CEO, Moonshots).
- Oct 16, 2026 5 new claims (All-In, Moonshots).
- Oct 15, 2026 1 new claim (All-In).
- Oct 11, 2026 David Sacks moved from “No, it mainly exhibits computational reasoning without leading to welfare breakthroughs.” to “AIs are solving Millennium Prize math problems.”.
- Oct 11, 2026 10 new claims (The Diary Of A CEO, All-In, Dwarkesh Podcast, Moonshots, 80,000 Hours Podcast).
- Oct 10, 2026 5 new claims (All-In, Dwarkesh Podcast, Moonshots).
- Oct 8, 2026 5 new claims (The Diary Of A CEO, Dwarkesh Podcast).
- Oct 4, 2026 10 new claims (The Diary Of A CEO, Doom Debates, Dwarkesh Podcast, 80,000 Hours Podcast).
Published because: 3 shows, 2 positions, 34 days of debate. Questions publish only when claims come from at least three shows, at least two positions are held by different people, and the debate has lasted two weeks.
Pairs are matched by content across different episodes and shows; no exchange between these speakers took place unless both quotes come from the same episode.
The prose on this page is written by a model from the quotes shown, and every sentence of it rests on quotes you can open; sentences that could not be traced to a quote were removed before publishing. Positions are placed from each speaker's own words. Evidence labels say what the speaker cited, never whether it is true.
4 episodes behind this page
- 2026-10-15Moonshots — OpenAI Bulk-Solves 722 Math Problems, Anthropic’s Cruelty Policy, and Elon’s Quadrillion-Dollar SpaceX | EP #302
- 2026-10-10All-In — Is Claude Conscious? Pope Rejects, Model Welfare Movement, OpenAI's Math Backlash, France Riots
- 2026-09-28Moonshots — Should we slow down AI progress? | MOONSHOTS #288
- 2026-09-11Dwarkesh Podcast — AI researchers debate how close we are to recursive self-improvement
The week's bottom lines, in your inbox
One email a week: the stories that mattered and what they actually mean. Free.