But in AI research, the priorities are exactly reversed. AI labs mainly care about how any new algorithmic recipe empirically makes training more efficient or elicits smarter models. They don't care about theory for its own sake, and all the progress we've made in deep learning over the last decade and a half has not required much new theory anyway. The equivalent of grinding through random conjectures would be enormously valuable in machine learning in a way that it just isn't in mathematics.
You might also like
Current agentic systems cannot predict the consequences of their actions. They apply recipes that have been trained to accomplish without really being able to predict the consequences. And I think it's a very bad way to build agentic systems, not a particularly safe way either.
Semafor
Mark Chen says OpenAI has built "AI research interns." So how far are we from AI that can do the research on its own?
The a16z Show
when AI solved some of its hardest problems: Navier-Stokes, the quasi-Riemann hypothesis, the unique games conjecture and hundreds of other problems.
Lex Fridman Podcast