Project ideas from Hacker News discussions.

Why I didn’t sign the Fields medallists’ letter

📝 Discussion Summary (Click to expand)

1. Concern that AI undermines deep mathematical understanding

Many commenters worry that AI's instant "true/false" results short-circuit the learning process, replacing deep insight with superficial answers.

"While LLMs are definitely capable of solving problems in search of truth, I agree with Tao that instant 'true/false' results threaten to short-circuit the traditional avenues we have used to escape local minima in the search for truth. Their products may be the junk food that provides immediate satiation in exchange for long-term health." – piker
"The letter is not incompatible with 'pro AI in math' [...] It’s primarily 'anti-unnecessarily-destroy-the-human-systems-of-mathematics-just-for-benchmarks-and-marketing'" – GPerson

2. Tension between adaptation versus resistance to AI

Debate persists over whether mathematicians should adapt to AI tools or push back against certain implementations, with arguments about community agency and coordination.

"Why can't OpenAI publish whatever it wants. And the math community can use it or not use it [...] I say that the community can rearrange and allocate prestige and time in different ways to maximally use the technology." – simianwords
"You're describing an improvised surgery on a living organism. Developing a complex system involving humans that is productive and doesn't collapse is extremely hard, so if it ain't broke don't fix it." – neuroticnews25

3. Misalignment of goals between AI companies and mathematicians

Frequent criticism that AI companies prioritize benchmarks, marketing, and investor returns over genuine mathematical progress and community values.

"...the push by AI companies to solve mathematical problems as a benchmark is detrimental to the science of mathematics, and to the mathematical community. The goals of the AI companies and the goals of the mathematical community are severely misaligned." – GPerson
"AI companies treat them like they treat absolutely everything else: natural resources, literature, art, code etc. as something to be chucked into the ravening maw and pooped out the back as profit." – fruitl00p

4. Value of human understanding versus mere problem-solving

Ongoing debate about whether mathematics is primarily about attaining conceptual understanding or solving problems, and whether AI threatens the former.

"However, math problems are really there to solve a real world problem. this is ABSOLUTELY not how actual mathematicians see their field. A problem in mathematics is just that: its interest to a pure mathematician is not related to any applications in other disciplines." – nicebyte
"Understanding the world and math, and discovering beautiful explanations is a worthy pursuit for humans in and of itself." – jgord

5. Impact on mathematical social structures and incentives

Concerns that AI could erode prestige systems, reduce incentives for deep work, disrupt funding, and damage the social processes that sustain mathematical progress.

"Thus, the primary risk, as I see it, is that a lot of people who would have done a PhD in mathematics and gone on to become custodians of the mathematical tradition will no longer wish to do so." – NonHyloMorph
"we urgently need to come up with good ways of explaining the value of having a large pool of human mathematical experts, even if it is no longer part of their role to find new proofs of theorems" – layer8 (quoting Gowers' blog)


🚀 Project Ideas

ProofDigest

Summary

  • A collaborative annotation platform where mathematicians can break down AI-generated proofs into human‑understandable steps, add intuition, and link to related concepts.
  • Turns opaque AI “true/false” outputs into learning resources, preserving the social process of mathematics and providing prestige for explanation work.

Details

Key Value
Target Audience Research mathematicians, graduate students, math enthusiasts
Core Feature Upload AI proof (text/Lean/Coq), annotate with natural‑language explanations, tag steps, link to definitions/variants, vote on clarity
Tech Stack React frontend, Node.js/Express backend, PostgreSQL, optional Lean 4 integration for proof checking
Difficulty Medium
Monetization Revenue-ready: Subscription for teams/institutions (SaaS) with free tier for individuals

Notes

  • HN commenters warned that “instant true/false results threaten to short‑circuit the traditional avenues we have used to escape local minima” and stressed the need to digest proofs. ProofDigest directly enables community‑driven digestion, addressing that pain point.
  • By awarding reputation for clear explanations, it creates a new prestige channel that aligns with the community’s desire to reward understanding, not just answer‑generation.

VeriProof Marketplace

Summary

  • A bounty‑based verification service where humans earn rewards for checking the correctness of AI‑generated proofs and producing human‑readable proof sketches or formalizations in proof assistants.
  • Provides financial incentive for humans to verify and explain AI proofs, aligning AI company goals with community values.

Details

Key Value
Target Audience Mathematicians, formal‑methods experts, proof‑assistant users
Core Feature AI proof submission, bounty placement, verification submission (informal explanation or formal proof), review, payout
Tech Stack Smart‑contract escrow (e.g., Polygon) or traditional escrow, React UI, backend in Rust/Go, integration with Lean/Coq
Difficulty High (due to blockchain/payments and proof‑assistant integration)
Monetization Revenue‑ready: Take a small fee (≈5%) from each completed bounty

Notes

  • Commenters noted that “the math community can rearrange and allocate prestige and time in different ways” and that incentives are needed to motivate engagement with solved problems. VeriProof offers tangible rewards for verification and explanation.
  • The marketplace could spark discussion on proper incentive structures and reduce the feeling that AI companies are “scooping” results without giving back to the community.

ConjectureCraft

Summary

  • An AI‑assisted tool that helps mathematicians generate interesting new conjectures and problem variations, preserving the problem‑finding aspect of mathematics that AI might otherwise diminish.
  • Augments human creativity, ensuring mathematicians remain upstream of problem generation rather than just consumers of AI solutions.

Details

Key Value
Target Audience Researchers in pure math, graduate students
Core Feature Input a theorem or area; AI suggests related open problems, conjectures, or variations based on literature mining and analogical reasoning; user can rank, refine, and save ideas
Tech Stack Python backend (open‑source LLM such as Llama 3), vector DB for paper embeddings, frontend in Svelte or Vue
Difficulty Medium
Monetization Hobby (free, open‑source; premium model access could be added later)

Notes

  • Discussion highlighted fears that AI will “remove the incentive to be in this field” by solving problems instantly, leaving no room for problem discovery. ConjectureCraft keeps humans in the loop for generating new questions, a point several users praised as essential for lasting progress.
  • HN users would likely appreciate a tool that augments rather than replaces creativity, echoing calls for “the community can rearrange and allocate prestige … to maximally use the technology.”

ProofUnderstanding StackExchange (MathProof.SE)

Summary

  • A specialized Q&A site focused on explaining AI‑generated proofs, where users ask “How does this proof work?” and experts provide intuitive explanations, connections to known techniques, and references.
  • Creates a reusable knowledge base of proof explanations, addressing the erosion of understanding and providing a venue for reputation building.

Details

Key Value
Target Audience Mathematicians, students, educators
Core Feature Question posting with AI proof attached, answers with upvotes, tags for topic, ability to link to formalizations, reputation system
Tech Stack Open‑source Q&A platform (e.g., Answer.dev or a Node.js/PostgreSQL clone), LaTeX rendering via MathJax
Difficulty Low‑Medium (leverages existing Q&A software)
Monetization Hobby

Read Later