Replies

• 

  John Carlos Baezjohncarlosbaez@mathstodon.xyz

  Sep 19, 2026, 9:12 PM

  @maxpool - It's interesting to read the comments on Tao's video announcing a new initiative to create an "open math model, built with the mathematical community: something researchers can inspect, run, teach with, and improve." Some are sympathetic, while some others aren't: "When Tao says to use AI responsibly what he really means is to use it in the way that he approves of, in a vain attempt to protect his institution as he sees it." Academics are usually somewhat insulated from these skeptics, but right now some mathematicians are very publicly struggling to cope with prooflation.

  The initiative itself is also interesting, of course!

  https://www.youtube.com/watch?v=PZRb6NIki2w

  💬 1🔄 0⭐ 0
• 

  ma𝕏poolmaxpool@mathstodon.xyz

  Sep 19, 2026, 11:21 PM

  RE: https://mastodon.social/@maxkasy/117299011082820470

  @johncarlosbaez

  I suspect this is the future.

  @maxkasy just made a relevant post. I think many mathematicians would opt to be part of a reinforcement learning loop if the model is open and developed in academia, and it will become a common resource.

  https://mathstodon.xyz/@maxkasy@mastodon.social/117299012390106577

  💬 0🔄 0⭐ 0
• 

  Σ(i³) = (Σi)²SvenGeier@mathstodon.xyz

  Sep 19, 2026, 11:15 PM

  @maxpool

  >shifting the primary bottleneck of mathematics from establishing truth to extracting insight

  Wasn't that always the primary bottleneck?

  💬 0🔄 0⭐ 0
• 

  SemionVSD98@mastodon.social

  Sep 20, 2026, 1:55 AM

  @maxpool There may be claims which are not necessarily of one's prime interest but one wishes just to know the truthness of. Something one can use as a lemma, for example. The full comprehension is not needed in the situation, so quick proofs from AI well fit in there

  💬 0🔄 0⭐ 0