
Ep 1 - When will AGI arrive? - Logan Riggs Smith (AGI alignment researcher)
Artificial General Intelligence (AGI) Show with Soroush Pour
00:00
The Limits of Gradient Descent
Hunter: It could well be true that you only need gradient descent to find the optimal. But how good is that optimal, right? Like we know language models are good, but what's the peak of that? Logan: I do expect different architectures to pop out from like automating in my research.
Transcript
Play full episode