saved
Why I didn’t sign the Fields medallists’ letter
Hraness cites a source capture. The source author remains the source.
gist
Timothy Gowers explains why he declined to sign the Fields medallists’ severe-misalignment letter on AI and mathematics, while agreeing a crisis exists and refusing an opposing camp. He rejects treating conceptual understanding as the sole primary goal over problem-solving, argues a flood of AI results can increase digested mathematics even as undigested work rises, and locates the real risk in social structures: fewer PhDs if problem-solving dreams vanish, and policy that stops funding human mathematical experts. Cross-posted to Terence Tao’s blog.
ideas
- Spectrum, not a single right temperament. Problem-solving and conceptual understanding sit on a spectrum; the letter’s framing risks declaring conceptual primacy the only correct mathematical attitude.
- Flood can raise digested output. Even if AI volume leaves more important work undigested, it can still increase the amount that is digested—a bargain if communities choose what to absorb in parallel.
- Citation failures look temporary. Missing human credit is serious now but the credit system itself may collapse once finding an amazing proof is like spotting a comet; hurry, not capability, may explain current sloppiness.
- Primary risk is succession and funding. The danger is fewer people entering PhDs and custodianship when famous-problem dreams fade, plus policymakers concluding mathematicians are no longer needed.
- Little to demand from companies. Near-term public models will flood results anyway; criticizing firms for generating solutions early may only bring forward the same transition by months.
quotes
“Perhaps the main reason I didn’t sign is that I don’t fully subscribe to this view.”
“the primary risk, as I see it, is that a lot of people who would have done a PhD in mathematics”
“a flood of “big” AI results would be likely to increase the amount of important mathematics that was not properly digested”
“I felt that there was nothing to be gained from criticizing AI companies for generating too many solutions too quickly.”