It goes beyond the title with direct discussion of like, it's, that's, including: Oh, I didn't sign into my own chat.
Key takeaways
01It's not doing like recursive rag search
02And so the reason that I'm like doing this and the reason that I'm kind of excited about it is because when you see a jump in benchmarks the way that we just saw, that basically me
03And when I say solves the problem, I mean solves reasoning and solves math, which means that language models, yes, are actually adequate to get at least 90%, like we're already clo
Best for
listeners looking for a practical AI episode debrief