An OpenAI model has disproved a central conjecture in discrete geometry 2 months ago
I see mixed emotions here. I understand both. On one hand it's exciting and fascinating. On the other it's concerning. One concern I haven't seen mentioned is the possibility that, as these models become larger and more powerful, their capability to solve frontier math problems will also grow. Does there become a point where humans are no longer the driver of innovation and research in this world, and instead are relegated to become stewards of the AI models whose purpose is to push the boundaries of mathematics, theoretical physics and other academic disciplines?