The other day a student asked me about the prevalence of insider trading in prediction markets. I now have an answer.
Happy Stanislav Yevgrafovich Petrov Saves the World Day, everybody.
42 years ago today, Lt. Col. Petrov prevented an accidental nuclear war.
en.wikipedia.org/wiki/Stanisl...
Two recent revisions have shown me how much LLMs have improved over the last year- Claude and Chat have both spotted errors in these two papers that neither they nor Refine spotted last year.
To be fair, none of the errors involved changing the assumptions of the results. It was just that some of the arguments were too loose.
Way to go Sally! links.mit.edu/nl3/N5Jy2cKq...
Or ignored edge cases, which actually is formally a mistake
Two thoughtful essays about the impact of LLMs on graduate education:
ergosphere.blog/posts/the-ma...
economics.mit.edu/sites/defaul...
From personal experience I think the self-control problem mentioned in the first essay is very real.
Is there currently a program that can determine whether a given matrix is completely positive? ChatGPT 4o can’t seem to do it even when asked to explain its reasoning.
Jason Furman
Elizabeth Bear
On AI agents, grunt work, and the part of science that isn't replaceable.
