3 ms·
Max Tegmark: The Case for Halting AI Development
- nh23423fefe 3y ago> AI is asteroid heading towards us and people are ignoring it > there is no middle ground between utopia/apocalypse outcomes > the original plan to develop "safety" isn't working so we need to stop developing capability > pause for 6 months to allow coordination But there is never a description of how safety comes to exist from pausing.
- ptrhvns 3y agoNo one is suggesting that we just pause and that's it. The pause is to give safety researchers time to do their work. Since progress is outpacing safety research, if we don't pause, we will eventually reach the cliff without a way to protect ourselves, for better or worse.
- nh23423fefe 3y agoWhy should i believe safety research will produce anything of value given any amount of time. The only thing we know for sure about safety research is that it reduces the performance of models. So it seems like a cost center just demanding more resources for no reason.
- ChatGTP 3y agoI guess you’ve benefited your whole life from safety research, engineers learning lessons and people doing things properly. I’m a bit confused why the rush anyway ?
- nh23423fefe 3y agoI've benefited my whole life from AI safety research? Who is rushing, that's just framing. AI safety invents a fake problem and fails to solve it and so demands that other productive people "slow down" to they can "catch up." See I can frame too.
- ChatGTP 3y agoI didn’t say that specifically about AI, but you likely will benefit from AI safety research. Anyway anyone born in the last century had benefited from lots of other regulations and safety researchers work, I hope this time it’s no different.