Tyler Cowen: belief

Transcript context

…To the extent that there was a trade off, in that hypothetical, would you put the same weight on extra work that they do? Or do you just differ with them on how you actually go about solving essential risks? Probably more the latter. I think they're not epistemically modest enough when it comes to existential risk. They think they have all these petite very particular hypotheses about AGI and we've got to prevent this. That's where I really differ from them. I think their ability to limit that risk, however great or small, might basically be zero. That if AGI is a risk, it's the worst set of procedures that will do you in and you can't regulate those very well at all. Putting everyone at your favorite tech company through this training about alignment… I'm not against doing that. But like, come on! You know, if it's gonna happen, it's like handling pandemic materials. It's the sloppiest people you've got to worry about, and they are not sitting in on your class on AGI in alignment. Yes. Although it isn't surprising to the extent that companies in the US that (maybe) care more than other other entities about alignment are actually the first currently. Like Open AI for instance.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.