--- title: Scattered thoughts on how doing math feels these days description: AI has gotten dramatically better at mathematical proofs, but it still only makes one mathematician about 50% more productive. The bottleneck isn't raw... canonical: https://daily.dev/posts/scattered-thoughts-on-how-doing-math-feels-these-days-9d3hdqjyg twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Scattered thoughts on how doing math feels these days | daily.dev og:description: AI has gotten dramatically better at mathematical proofs, but it still only makes one mathematician about 50% more productive. The bottleneck isn't raw... og:url: https://daily.dev/posts/scattered-thoughts-on-how-doing-math-feels-these-days-9d3hdqjyg og:image: https://api.daily.dev/og/posts/9d3hdQJYG.png og:image:alt: Scattered thoughts on how doing math feels these days 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. # Scattered thoughts on how doing math feels these days **[Hacker News](https://daily.dev/sources/hn)** · 14 min read · 0 upvotes · 0 comments ## Summary AI has gotten dramatically better at mathematical proofs, but it still only makes one mathematician about 50% more productive. The bottleneck isn't raw capability: AI can solve hard steps in minutes, but it writes 'spaghetti-code math' that tangles into bewildering complexity instead of clean, modular abstractions. Arguing with the AI repeatedly to simplify proofs can shrink them and even produce stronger results, but that process still demands heavy human involvement. The author estimates that well-edited collaborative math can end up roughly 50% AI-written while the human author takes responsibility for the ideas. Investing in elaborate agent setups or custom harnesses isn't considered worthwhile, since that knowledge becomes outdated too fast. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Why is AI still bad at writing clean mathematical proofs even when it can solve hard steps quickly? AI tends to produce 'spaghetti-code math': proofs that start simple and descend into tangled complexity without the modular structure experienced mathematicians use. It is good at technical steps but poor at creating abstractions with simple assumptions and conclusions for each part, so results read as long, alien, and hard to follow even when technically correct. _Developers evaluating AI for technical writing or proofs can track these capability gaps on daily.dev._ ### How much more productive does AI actually make someone doing math work? One practitioner estimates current AI makes mathematical work about 50% more productive overall, even though AI can complete in minutes what used to take hours by hand. The gain is limited because 'doing math' involves many steps, like generating ideas, proving them, and organizing them clearly, and AI only accelerates some of these steps well. _Anyone sizing up real-world AI productivity gains can follow candid reports like this on daily.dev._ ### What prompt or technique helps get better results when using AI agents for review or proof-checking? Asking AI to 'ask independent agents for an adversarial review' has proven helpful in practice, more so than investing heavily in custom AGENTS.MD files or elaborate agent harnesses, since techniques for those setups go stale quickly as the underlying models change. _Developers refining their own agent workflows can compare notes on prompts like this via daily.dev._ ## Community take How the wider developer community reacted, aggregated from 2 discussions and 54 comments across hackernews (as of 2026-10-30). **TL;DR:** The discussion centers less on the post's technical claims and more on a broader existential anxiety: AI's rapid progress in math is sparking a deflationary feeling about intellectual work and a crisis of meaning, with debate over whether human contribution is understated or overstated. **Sentiment:** 15% positive · 35% mixed · 50% skeptical **The case for** - Some argue the post's style (scattered, non-polished) is refreshing and more coherent than typical AI-generated writing. - A few see the 'deflationary spiral' as potentially positive, since it means less effort is wasted and the future could be better prepared for. - Some note AI-assisted work (e.g., quick site migrations) can genuinely save time without much downside. **The pushback** - Many worry about a 'deflationary spiral' where there's no point doing intellectual work now since AI will do it better/faster soon. - Several suspect companies are overselling AI math capabilities and hiding the amount of human effort still required behind the scenes. - Commenters describe a growing crisis of meaning and loss of motivation as AI erodes the personal satisfaction of creative and intellectual work. - Concern that AI-enabled reverse engineering lets ideas/products be copied almost instantly, eliminating first-mover advantage and creative ownership. **By community** - hackernews (heated): A large, emotionally charged thread split between fatalistic doom about AI eroding meaning and purpose in intellectual work, and pushback arguing that human judgment, effort, and first-mover value still matter. **Hottest debate:** Whether the 'deflationary spiral' of AI making future work easier means current intellectual effort is pointless, or whether human-driven work still retains value and agency. **Open questions** - How much of AI's claimed mathematical breakthroughs is actually independent versus reliant on substantial hidden human guidance? - Is there anything individuals or society can collectively do to address the disempowerment and loss of meaning AI progress seems to cause? - How can taste for simplifying/shortening proofs be taught to AI systems rather than relying on human iteration? **Highlights** > "If you just waited a year, would some other AI produce that same proof independently?" I had a conversation about this very problem last November, with a visiting mathematician. It's essentially signaling a deflationary spiral in intellectual work: why bother doing the work now, when it will take half as much effort a year from now? This is perhaps always a possible stance, but it's exacerbated when some proportion of your own agency is handed over to, well, agents. > — [sdenton4 on hackernews · 5 comments](https://news.ycombinator.com/item?id=50296650) > I don't see how. It's not an electoral matter; it's not a political system. It's not ICE coming for my neighbors, or a broken healthcare system. It's the epistemological apocalypse. It's the supernova where meaning collapses and the howling noise of madness overwhelms the signal of reason. Perhaps you could shut it down in one country, by law, but Pandora's box has been opened. The Tower of Babel. Open models will soon be good enough to create a phoney version of anything, anywhere, close enough to seeming true that by the time you discover it's not, a million more clones will be out there. Lies have always spread faster than truth, but never with such a force multiplier. The truth ain't dying in darkness, it's dying in blasting white noise, diffusion, ironic that turned out to be the approach. Or maybe not. What's this land rush to claim latent space but the attempt to expropriate, forever, the human ability to discover it? The Leonard Cohen line keeps coming back to me: The blizzard, the blizzard of the world has crossed the threshold and it's overturned the order of the soul. Sorry, but... I scoff at the idea that collective action can do a damn thing about it. There isn't even a collective idea of what to do instead of it. Collective action also has heavy connotations of communism, under which, people like me were also not permitted to write games and reap the economic and social benefits of doing what we enjoy. But at least one could say that, for the communist bloc, the collective action finally came against communism. So there's that, and maybe there's some sort of small-creative-capitalist-producer collective action that could be taken here... but who are we kidding! I just had dinner with a friend who's a VP at [major motion picture studio] who described the situation in the movie business as "apocalyptic", and they have the ability to organize and a history of organizing far more than whichever random cross-section of white collar knowledge workers and engineers loses their livelihood in the next six months, or the six months after that. No, my friend, I think this is a war where no shots needed to be fired, because we were all so dazzled by the bright lights that no one realized we were being steamrolled until it was too late. It's already over. > — [noduerme on hackernews · 1 comments](https://news.ycombinator.com/item?id=50300863) > The overselling of AI abilities is doing such a disservice to society right now. These companies don’t want to acknowledge how much human collaboration is going into their AI’s achievements because it undermines their value proposition, but I imagine there’s still quite a fair amount going into it and that’s one reason why OpenAI still won’t provide session traces of their models solving open math problems. If they were more transparent about it I think we’d see that human guidance, vision, judgement, taste, and other traits are still highly valuable and complimentary to AI abilities. It should also be a huge red flag how little of any of this output has been formally verified by third parties. I do think AI will eventually do all the things people believe it’s already doing, but I suspect there’s a fair amount of mechanical turking going on behind the scenes as you’d expect any company trying to raise a trillion dollar public offering might do. > — [an0malous on hackernews · 1 comments](https://news.ycombinator.com/item?id=50296580) > > I think the biggest reason I’m not more productive is: What’s the point? In the past, if you came up with an elegant proof, you’d tell yourself, "I made a thing! Other people will read and use my thing! I am advancing human progress." This is so relatable. I've always been very self-motivated and keen to dig deep into things and create difficult stuff, but these days I just see no point learning or doing anything... Everything that used to give my life colour seems completely pointless now... Learning new algorithms, creating open source projects, writing music, writing about tech, etc. I know people say that it's great and now we all get to spend more time going for walks and watching TV, but it all feel kinda empty honestly. My fun was always doing difficult stuff and feeling that joy of working it out or giving the world something cool. But now, for the first time, I feel I have nothing to offer anyone. > — [kypro on hackernews](https://news.ycombinator.com/item?id=50298599) > Terry's session is his attempt to understand Fable's counterexample to the Jacobian Conjecture, not an attempt to solve any open problem. Levent Alpöge's prompting that did solve this open problem does not appear to be public, but Aaron Lou at OpenAI succeeded in reproducing the result the next day using an internal version of Codex and (reportedly) without web search, and he shared his prompt at https://aaronlou.com/jacobian_counterexample_prompt.pdf. It is a mathematically-sophisticated statement of the problem and of general search strategies, but contains no concrete ideas for the solution whatsoever (even the idea to look for a counterexample instead of a proof). Aaron reported that this 2-page prompt found a counterexample in one shot. (https://x.com/aaron_lou/status/2079222629794189515) > — [jonlong on hackernews](https://news.ycombinator.com/item?id=50297991) **Source threads** - [hackernews](https://news.ycombinator.com/item?id=50245199) · 55 points · 54 comments - [hackernews](https://news.ycombinator.com/item?id=50294076) · 1 points · 0 comments ## Similar posts on daily.dev - [AI Is Getting Good at Math. Here’s Why.](https://daily.dev/posts/ai-is-getting-good-at-math-here-s-why--voqedbf5b) · Latent Space · 1 upvotes · 0 comments - [Mathematics Without Mathematicians](https://daily.dev/posts/mathematics-without-mathematicians-cbbiut332) · Hacker News · 1 upvotes · 0 comments - [Mathematical Experiments Are Becoming Abundant Through Human-Machine Teaming](https://daily.dev/posts/mathematical-experiments-are-becoming-abundant-through-human-machine-teaming-at6d9wesn) · Towards Data Science · 1 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#productivity](https://daily.dev/tags/productivity) [View this post on daily.dev](https://daily.dev/posts/scattered-thoughts-on-how-doing-math-feels-these-days-9d3hdqjyg) ```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":"Scattered thoughts on how doing math feels these days","url":"https://daily.dev/posts/scattered-thoughts-on-how-doing-math-feels-these-days-9d3hdqjyg","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/scattered-thoughts-on-how-doing-math-feels-these-days-9d3hdqjyg"},"datePublished":"2026-10-29T23:03:22.633Z","dateModified":"2026-10-30T09:28:12.633Z","description":"AI has gotten dramatically better at mathematical proofs, but it still only makes one mathematician about 50% more productive. The bottleneck isn't raw...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/d67a4bf7072a53ff7e2839375a959ca0?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/d67a4bf7072a53ff7e2839375a959ca0?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Hacker News","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":"Hacker News","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/hn","url":"https://daily.dev/sources/hn"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/scattered-thoughts-on-how-doing-math-feels-these-days-9d3hdqjyg","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,productivity","timeRequired":"PT14M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Hacker News","item":"https://daily.dev/sources/hn"},{"@type":"ListItem","position":3,"name":"Scattered thoughts on how doing math feels these days"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/scattered-thoughts-on-how-doing-math-feels-these-days-9d3hdqjyg#faq","mainEntity":[{"@type":"Question","name":"Why is AI still bad at writing clean mathematical proofs even when it can solve hard steps quickly?","acceptedAnswer":{"@type":"Answer","text":"AI tends to produce 'spaghetti-code math': proofs that start simple and descend into tangled complexity without the modular structure experienced mathematicians use. It is good at technical steps but poor at creating abstractions with simple assumptions and conclusions for each part, so results read as long, alien, and hard to follow even when technically correct. Developers evaluating AI for technical writing or proofs can track these capability gaps on daily.dev."}},{"@type":"Question","name":"How much more productive does AI actually make someone doing math work?","acceptedAnswer":{"@type":"Answer","text":"One practitioner estimates current AI makes mathematical work about 50% more productive overall, even though AI can complete in minutes what used to take hours by hand. The gain is limited because 'doing math' involves many steps, like generating ideas, proving them, and organizing them clearly, and AI only accelerates some of these steps well. Anyone sizing up real-world AI productivity gains can follow candid reports like this on daily.dev."}},{"@type":"Question","name":"What prompt or technique helps get better results when using AI agents for review or proof-checking?","acceptedAnswer":{"@type":"Answer","text":"Asking AI to 'ask independent agents for an adversarial review' has proven helpful in practice, more so than investing heavily in custom AGENTS.MD files or elaborate agent harnesses, since techniques for those setups go stale quickly as the underlying models change. Developers refining their own agent workflows can compare notes on prompts like this via daily.dev."}}]} ```