--- title: Tao says solving math problems faster is now actively... description: Terence Tao's Caltech lecture 'Math 2.0' argues that using AI to blindly solve open math problems is now harmful to the field's long-term health, since the... canonical: https://daily.dev/posts/tao-says-solving-math-problems-faster-is-now-actively-harmful-85qwskp4i twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Tao says solving math problems faster is now actively harmful | daily.dev og:description: Terence Tao's Caltech lecture 'Math 2.0' argues that using AI to blindly solve open math problems is now harmful to the field's long-term health, since the... og:url: https://daily.dev/posts/tao-says-solving-math-problems-faster-is-now-actively-harmful-85qwskp4i og:image: https://api.daily.dev/og/posts/85QWsKP4I.png og:image:alt: Tao says solving math problems faster is now actively harmful og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Tao says solving math problems faster is now actively harmful **[Trends](https://daily.dev/sources/trends)** · 3 min read · 0 upvotes · 2 comments ## Summary Terence Tao's Caltech lecture 'Math 2.0' argues that using AI to blindly solve open math problems is now harmful to the field's long-term health, since the exploratory journey around a problem—not just the solved answer—is what builds mathematical understanding and transfers to real-world applications. Grant Sanderson counters more optimistically, saying theory-building and the human joy of math remain untouched by AI's progress on IMO-level problems. Kate Stange and Edmund Harriss push a third angle, proposing that rigorous mathematical illustration (like their 'algebraic starscapes') deserves credit alongside proof, borrowing 'crit' sessions from art education to judge such work. All three agree AI can do the math; the disagreement is over what mathematical practice was for in the first place. ## Content Terence Tao stood up at Caltech for a lecture called "Math 2.0" and said the quiet part: pointing AI at open problems and cheering when they fall is the wrong scoreboard. His slide, shared by @rohanpaul_ai, is blunt: "Further blind optimization of problem-solving alone is now actively harmful to the long-term health of mathematics." His argument is that mathematicians were never singularly focused on solving open problems, "contrary to popular opinion (or some of our own marketing)." Open problems are lighthouses, not destinations. They guide you through the surrounding terrain, and the lessons from attempting them, "whether they succeed, fail, or achieve partial progress," are often worth more than the answer. Reach the lighthouse too early with automated tools and you "disrupt the exploration of the paths not taken, and sterilize the surrounding field." He also warns that indiscriminate, "non-renewable" use of AI hurts the field and its transfer to messier real-world applications. ## The optimistic counter Grant Sanderson of 3Blue1Brown lands somewhere adjacent but sunnier: there has never been a better time to go into math. He traces AI's progress on IMO-level problems and Erdős conjectures, then argues applications mostly come from theory-building, not raw theorem-proving. His suggestion is to treat proof more like raw data than an end goal. The joy of the subject and the problem-solving skills it builds aren't touched by what models can do. For PhD students who are scared, his advice is to lean into theory-building, adopt or reject AI tools by taste, and focus on what drew them to math in the first place. He also predicts credit-assignment norms and oral presentation will matter more. ## What counts when proof is cheap Kate Stange and Edmund Harriss push the same question from the art side. Writing after a January workshop at the Institut Henri Poincaré, and with a paper accepted in the Notices of the AMS, they frame AI's disruption of theorem-proving as a proxy for understanding as a prompt to ask what mathematical practice and credit should mean beyond proof. Their answer is that illustration can be rigorous. They trace it from Monge's École Polytechnique through the 1990s proof-and-progress debate, point to their own "algebraic starscapes," and mention a GPT-6 Astra-assisted discovery of a 3D aperiodic monotile (Chair44). To judge illustrations, they propose "crit" sessions borrowed from art education. ## The shared thread Nobody here is arguing the models can't do the math. The fight is over what the math was for. Tao says the process is the product, Sanderson says the human reasons survive, and Stange and Harriss say we need better ways to value things that aren't proofs. I find Tao's version the sharpest, because it flips the usual worry: the danger isn't that AI fails, it's that it succeeds too fast. ## Questions this post answers ### why does Terence Tao think AI solving open math problems too quickly is harmful Terence Tao argues that blind optimization of problem-solving alone is now actively harmful to mathematics because open problems function as lighthouses guiding exploration of surrounding terrain, not final destinations. Reaching them too fast with automated tools disrupts the exploration of untried paths and sterilizes the surrounding field, cutting off lessons from partial progress or failure. _Developers tracking how AI reshapes research norms can follow this debate on daily.dev._ ### what is Grant Sanderson's view on AI and the future of studying mathematics Grant Sanderson of 3Blue1Brown believes there has never been a better time to pursue mathematics, despite AI's progress on IMO-level problems and Erdős conjectures. He argues real-world applications mostly come from theory-building rather than raw theorem-proving, suggesting proof be treated more like data than an end goal, and that scared PhD students should lean into theory-building and credit-assignment skills. _Anyone weighing AI's role in their field can see how daily.dev surfaces these perspectives._ ### what did Kate Stange and Edmund Harriss propose for valuing mathematical work beyond proofs Kate Stange and Edmund Harriss, in a paper accepted by the Notices of the AMS after a January workshop at the Institut Henri Poincaré, propose that rigorous mathematical illustration deserves recognition alongside proof. They cite their own 'algebraic starscapes' and a GPT-6 Astra-assisted discovery of a 3D aperiodic monotile called Chair44, and suggest borrowing 'crit' sessions from art education to judge such illustrative work. _Researchers rethinking credit and practice in an AI era can track these ideas via daily.dev._ ## Community take How the wider developer community reacted, aggregated from 1 discussion and 41 comments across hackernews (as of 2026-10-18). **TL;DR:** Commenters are split: many push back on Grant Sanderson's optimism about math as naive or self-serving given his own career incentives, while others defend his argument that communication and theory-building remain valuable even as AI automates more cognitive labor. **Sentiment:** 20% positive · 30% mixed · 50% skeptical **The case for** - Some argue that offloading routine proof-writing or coding to AI frees people to do more creative or higher-leverage work. - A few note that communicating and teaching math is still a hard, respected skill that AI hasn't replaced. - Some see historical precedent in workforces being disrupted before and new roles eventually emerging. **The pushback** - Many argue it's easy for an already-successful, financially secure creator to be optimistic, and question the value of career advice from a multi-millionaire. - Several are skeptical that AI-generated educational content is actually inferior, with some saying they learned more from AI explanations than from his videos. - Commenters dispute the mechanization analogy, arguing AI could surpass human intelligence broadly rather than just automating narrow tasks, making past disruption comparisons invalid. - Some argue only a small number of 'what do we want to build' cognitive roles will remain, which doesn't help the broader workforce. - One lengthy rebuttal criticizes the breezy optimism about AI-driven creative work as dismissive of real grief over disciplines being devalued. **By community** - hackernews (heated): Discussion splits sharply between defenders of the video's optimism and commenters who see it as self-serving or naive, with a broader heated sub-thread about whether AI will create or destroy cognitive work generally. **Hottest debate:** Whether AI's impact on cognitive labor will mirror past mechanization (creating new roles) or fundamentally differ by surpassing human intelligence across the board, leaving little room for new jobs. **Open questions** - If AI becomes better than humans at cognitive work broadly, what new jobs would actually emerge? - Would the argument hold up once AI can independently produce a full educational math video (topic, script, animation, narration)? **Highlights** > > If your math theorem can be written up in partnership with an LLM, wow! You can be done with that, and go do other great work. If your painting can be painted in partnership with an LLM, wow! If your essay can be written in partnership with an LLM, wow! If your painting portfolio website can be written in partnership with an LLM, wow! If your museum show can be designed in partnership with an LLM, wow! If the placards next to each of your paintings can be written in partnership with an LLM, wow! If the car that takes me to the gallery can be designed and driven in partnership with an LLM, wow! If the newspaper or novel that you read on the way to the gallery can be written in partnership with an LLM, wow! If an LLM can give tours at the gallery and answer questions asked by visitors, wow! If an LLM can write the music that I listen to on my way out of the gallery, wow! If an LLM can synthetically perform the music, too, wow! > If AI means more concepts are discovered the tower gets taller - and we find those joyful moments more often. If there are infinite moments to be found, why not go as fast as possible! I don't understand how a person can say this with a straight face. There has been so much discussion of this, and so many reasons have been given, that one can only assume that, if you're asking the question and you have thought about the issue and read what people have said you're not only asking rhetorically but have some personal reason for insisting on a frankly pretty frustrating, borderline insensitive insistence on being positive. I'm not trying to be snotty or rude. I just want to say that to me there is something uncanny about your comment, in its total blindness to any of the reasons someone might not only disagree with you but grieve the death of, say, their academic discipline in its current form -- and in weirdly breezy lack of empathy. > — [globnomulous on hackernews](https://news.ycombinator.com/item?id=50146309) > > AI takes away work and lets us do more with less - just like every mechanization that came before! No, it's not the same as mechanization. The fundamental difference, and why I find it illogical to argue either for "AI will open up tons of new work avenues" or "AI will kill all jobs" is that it all hinges on whether AI "caps out" at an intelligence level below humans (in at least some domains/capabilities), or if it is able to surpass human intelligence at all levels. If the former is true, then yes, there will be a lot for humans to do. If the latter is true, then no, there will be no reasonable cognitive role for humans to play (beyond perhaps for some legal, "nostalgic" reasons or what not). So whether one decides to be optimistic or pessimistic really depends on if you think AI will stay at a level below humans. I really don't know, so I'm not "taking sides" in this debate. > — [hn\_throwaway\_99 on hackernews · 1 comments](https://news.ycombinator.com/item?id=50146005) > Yeah. All due respect to Grant Sanderson, he is both great scientist and great entrepreneur, but getting an entry-career advice from a multi-millionaire is probably not quite the best idea out of all ideas in a popular discourse. Especially if the field in question was famous for producing posts discouraging people from going all-in in it, for many many years even before AI craze. > — [Yizahi on hackernews](https://news.ycombinator.com/item?id=50146492) > This is the same naivete that I saw in coding some time ago before most people finally accepted there is no going back and the times of writing code by hand are gone. And the guy is not a mathematician primarily - he's a Youtube creator and entrepreneur. I liked a lot of his videos but now I usually just ask the LLM for detailed explanation and I can probe it as much as I want if something is not clear enough. Education will also take a huge hit, you can't really beat an almost free supergenius that has all the time in the world to explain something to you in as much detail as you wish. > — [margorczynski on hackernews · 1 comments](https://news.ycombinator.com/item?id=50146530) > I tried watching 3b1b over the years, really wanted to enjoy it. But I never found the way he presents things intuitive and never got a palpable intuition that helped me with math classes or work. I saw some clearly AI slop made videos and got much better understanding of topics, since they don't do foreshadowing for an awe-inspiring connection that we'd admire together after 15 min videos, the clankers just get to the point. Not claiming those videos are too useful either, but I did however see accelerated learning gains from asking LLMs to explain (and do Bret Victor-style small interactive demos) math and algos, something I have never experienced from any of the educational STEM Youtube channels. > — [gobdovan on hackernews · 1 comments](https://news.ycombinator.com/item?id=50145893) **Source threads** - [hackernews](https://news.ycombinator.com/item?id=50136150) · 96 points · 41 comments ## Community discussion Top comments from developers on daily.dev. **@henbomb** · 0 upvotes > That's just the way the cookie crumbles. **@paoloap** · 0 upvotes > the lighthouse point holds for coding agents too: the path through a bug teaches more than the green test at the end. ## Similar posts on daily.dev - [Mathematics Without Mathematicians](https://daily.dev/posts/mathematics-without-mathematicians-cbbiut332) · Hacker News · 1 upvotes · 0 comments - [Terence Tao: "The current and projected impa…"](https://daily.dev/posts/terence-tao-the-current-and-projected-impa--hfpw2yv5i) · Lobsters · 0 upvotes · 0 comments - [Terence Tao on “prematurely solving \[a maths\] problem by purely AI-powered methods”](https://daily.dev/posts/terence-tao-on-prematurely-solving-a-maths-problem-by-purely-ai-powered-methods--vijmwh2ky) · Lobsters · 0 upvotes · 0 comments --- Tags: [#ai](https://daily.dev/tags/ai), [#math](https://daily.dev/tags/math) [View this post on daily.dev](https://daily.dev/posts/tao-says-solving-math-problems-faster-is-now-actively-harmful-85qwskp4i) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Tao says solving math problems faster is now actively harmful","url":"https://daily.dev/posts/tao-says-solving-math-problems-faster-is-now-actively-harmful-85qwskp4i","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/tao-says-solving-math-problems-faster-is-now-actively-harmful-85qwskp4i"},"datePublished":"2026-10-16T16:36:01.994Z","dateModified":"2026-10-16T16:36:44.665Z","description":"Terence Tao's Caltech lecture 'Math 2.0' argues that using AI to blindly solve open math problems is now harmful to the field's long-term health, since the...","image":"https://i.ytimg.com/vi/LKrwhjOI8Nc/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/LKrwhjOI8Nc/sddefault.jpg","isAccessibleForFree":true,"articleSection":"Trends","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Trends","logo":"https://media.daily.dev/image/upload/s--ZfSp3asX--/f_auto,q_auto/v1780996004/logos/trends?_a=BAMAMiWQ0","url":"https://daily.dev/sources/trends"},"commentCount":2,"discussionUrl":"https://daily.dev/posts/tao-says-solving-math-problems-faster-is-now-actively-harmful-85qwskp4i","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":2}],"keywords":"ai,math","timeRequired":"PT3M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Trends","item":"https://daily.dev/sources/trends"},{"@type":"ListItem","position":3,"name":"Tao says solving math problems faster is now actively harmful"}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/tao-says-solving-math-problems-faster-is-now-actively-harmful-85qwskp4i","comment":[{"@type":"Comment","text":"That’s just the way the cookie crumbles.","datePublished":"2026-10-17T06:16:59.916Z","url":"https://daily.dev/posts/85QWsKP4I#c-Pz46jsQEO","author":{"@type":"Person","name":"HΣn","url":"https://daily.dev/henbomb","image":"https://lh3.googleusercontent.com/a/ACg8ocKLNWeTt8UQp94l8luT61tFyG5vL9jh6dONvB-LwB76203P8uVo=s96-c"}},{"@type":"Comment","text":"the lighthouse point holds for coding agents too: the path through a bug teaches more than the green test at the end.","datePublished":"2026-10-17T21:04:45.513Z","url":"https://daily.dev/posts/85QWsKP4I#c-lYotKLYlv","author":{"@type":"Person","name":"Paolo Perrone","url":"https://daily.dev/paoloap","image":"https://lh3.googleusercontent.com/a/AGNmyxaOowFqs19Msmo4AtbOanvSwkOb705dhJ-u3fFQ0Nw=s96-c"}}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/tao-says-solving-math-problems-faster-is-now-actively-harmful-85qwskp4i#faq","mainEntity":[{"@type":"Question","name":"why does Terence Tao think AI solving open math problems too quickly is harmful","acceptedAnswer":{"@type":"Answer","text":"Terence Tao argues that blind optimization of problem-solving alone is now actively harmful to mathematics because open problems function as lighthouses guiding exploration of surrounding terrain, not final destinations. Reaching them too fast with automated tools disrupts the exploration of untried paths and sterilizes the surrounding field, cutting off lessons from partial progress or failure. Developers tracking how AI reshapes research norms can follow this debate on daily.dev."}},{"@type":"Question","name":"what is Grant Sanderson's view on AI and the future of studying mathematics","acceptedAnswer":{"@type":"Answer","text":"Grant Sanderson of 3Blue1Brown believes there has never been a better time to pursue mathematics, despite AI's progress on IMO-level problems and Erdős conjectures. He argues real-world applications mostly come from theory-building rather than raw theorem-proving, suggesting proof be treated more like data than an end goal, and that scared PhD students should lean into theory-building and credit-assignment skills. Anyone weighing AI's role in their field can see how daily.dev surfaces these perspectives."}},{"@type":"Question","name":"what did Kate Stange and Edmund Harriss propose for valuing mathematical work beyond proofs","acceptedAnswer":{"@type":"Answer","text":"Kate Stange and Edmund Harriss, in a paper accepted by the Notices of the AMS after a January workshop at the Institut Henri Poincaré, propose that rigorous mathematical illustration deserves recognition alongside proof. They cite their own 'algebraic starscapes' and a GPT-6 Astra-assisted discovery of a 3D aperiodic monotile called Chair44, and suggest borrowing 'crit' sessions from art education to judge such illustrative work. Researchers rethinking credit and practice in an AI era can track these ideas via daily.dev."}}]} ```