Did you know about the big controversy in September in the mathematics field? OpenAI claimed its AI agents solved part of the million-dollar Millennium Prize math problem - the Navier-Stokes problem. There were rumours of an imminent human breakthrough by Tristan Buckmaster and Levent Alpogue. They were actively working on a solution to the problem and considered themselves months away from a solution. All of a sudden OpenAI puts 10,000 AI agents onto the problem and comes up with a solution in 88 hours.
Buckmaster "accused" OpenAI of focusing attention and speeding up activity on the problem. What he also worried about was that they could have stolen from the human researchers who had been using OpenAI as an aid in the research and storing results there. While OpenAI denies it would have stolen research directly, the researchers realized that they didn't opt out of OpenAI having access to their prompts as training input.
There's a frustration with AI companies treating important fundamental mathematical research as a media tool. This is the sort of thing that increases share value. Attention is distracted from the hacking, run-amok stories that make the headlines. Supposedly OpenAI wasn't going to claim the million dollars. I wonder why not? Why bother when they are about to hit the trillion dollar mark in corporate value and preparing for a "flotation" that could get it to the trillion dollar level.
The Mathematics community was very upset. An Open Letter was published by 25 Fields Medal winners who worry that the genuine pursuit of understanding is getting brushed aside in the pursuit of corporate victories. Having "the answer" isn't the same as showing how the answer is figured out. And to what else can be learned by studying it.
No comments:
Post a Comment