#49 - AGI: Could The End Be Nigh? (With Rosie Campbell)
Increments
00:00
The Most Promising Sub Areas of Safety
Vayne: I think there's a lot of interesting different directions to go in and I'm like very glad that people are pursuing these things. Reinforcement learning from human feedback, I think is one of these weird ones where it's like both a capability enhancer and potentially a safety enhancer but it has some pretty significant flaws when you consider it as the only safety enhancement. Vayne: There's probably a bunch of avenues that even if you're not sure I'm going to pan out they're still probably worth pursuing. But yeah, would I don't think we have anything that's like a perfect solution yet.
Transcript
Play full episode
