Post by @nilesh@fosstodon.org

2026-09-08 23:36 UTC

@drScott@mathstodon.xyz @tao@mathstodon.xyz I need to ponder more on this. I was reasoning under the scenario where most people have equal access to frontier AI. So, using AI becomes akin to using a calculator or a personal computer. But currently, it's definitely unequal (OpenAI reportedly spent more than $1m for Navier-Stokes), which is where "scooping" becomes possible, and might encourage not sharing promising research directions in the first place.

Replies (2)

@drScott@mathstodon.xyz 2026-09-09 03:06

@nilesh@fosstodon.org @tao@mathstodon.xyz I'm seeing reporting of $15m ( https://www.newscientist.com/article/2588063-openai-has-solved-the-navier-stokes-millennium-problem-using-15m-of-ai-effort/ ), though I haven't had time to check the sources. A wild amount of money to burn in a weekend.

Open ##5167471
@fenjen@mathstodon.xyz 2026-09-09 14:52

@nilesh@fosstodon.org @drScott@mathstodon.xyz @tao@mathstodon.xyz I feel you're missing one more reason that I do feel Tao's post touched on. It's not a formal reason, but more of a practical one: The system of mathematicians just practically doesn't really explore the search space of dead end paths after a path to a solution already has been found. The reasons for this can be multiple, perhaps incentive structures at an institutional level, or perhaps personal motivations not aligning or otherwise. Tao argue that this process is necessary, however, for important discovery and understanding. This is my understanding at least.

Open ##5167473