13 - First Principles of AGI Safety with Richard Ngo
AXRP - the AI X-risk Research Podcast
00:00
Is Scaling Up Neural Networks a Good Idea?
M cnu: I think there are a couple of corps intuitions which seem valuable. We're going to take it as sort of, for this discussion, that we should expect to get a g i at some point. M cnu: As you make relatively few big algrotmic changes, then it seems more plausible that actually you don't need a big paratime shift in order to reach a g i. You can just keep scaling things up, because that actually has worked last 70 years or so.
Transcript
Play full episode
