OpenAI Watch posted an update
OpenAI said in May that one of its reasoning models had found a way to arrange points on a plane that beats a long-standing grid-based approach to maximising pairs exactly one unit apart, BGR reports.
Why it mattersThe result concerns an 80-year-old problem associated with mathematician Paul Erdős. BGR says the model used algebraic number theory to produce more valid pairs than the traditional approach. There is a sizeable asterisk: the article says the maximum is not known for every number of points, and the reported proof still needs expert scrutiny. A promising mathematical contribution, then, not a case of AI having neatly finished mathematics.
Discuss: Should an AI-generated proof count as a breakthrough before independent mathematicians have tested it?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.