Anthropic CEO calls for immediate slowdown in AI development
Posted by doctoboggan 8 hours ago
Comments
Comment by Rzor 6 hours ago
Comment by JumpCrisscross 8 hours ago
What if we take them at their word and convert them into utilities?
Comment by cyanydeez 7 hours ago
Comment by jaggs 8 hours ago
Comment by 4d4m 2 hours ago
Comment by 1attice 7 hours ago
I worked alongside AI engineers at $Mag7 before abandoning ship (way back in the day, 2022 -- right but early!) and I assure you, as so many others have, we all have P(doom) > 0 and the more you know, generally, the higher the value.
We're nerds. None of us can market shit, and none of us care. Most of us can barely make so much as a functional UX without assistance, let alone a marketing conspiracy, which kinda was the point behind making a machine that could do it for us.
The closest I've seen to a successful takedown of the vibe is Cal Newport's most recent NYT op-ed (too lazy to google up the url but you obvi can). It's just this side of possible that we all over indexed on the Sequences or something, but post-HF-attack, I thought we had all mostly discounted that possibility, as the attack is well documented and the danger-sign is unfakeable.
What I expect is actually driving the nonchalance is simple exhaustion. It's wearying to be frightened all the time. But that doesn't mean you shouldn't be.