Quite the hot take from Alexey Guzey, that alignment is the bottleneck to capabilities and our alignment progress is going fine, so perhaps we should pause alignment to avoid speeding up capabilities? Yes, alignment techniques are dual use, but no, differentially only doing single-use stuff instead would not end better.
I feel like either Alexey is trying to gaslight people, or he interprets the terms “terrifying” and “an existential threat” to mean something quite different to what most people do. The latter feels faintly ridiculous. But only faintly. So much of my life has been wasted due to translational friction that my prior for it being the the source of disagreements is quite high.
If it wasn’t Guzey I would have dismissed the whole thing as trolling or gaslighting, and I wouldn’t have covered it beyond one line and a link. He’s definitely very confused somewhere.
For what it’s worth, he has shared (confidential) AI predictions with me, and I was impressed by just how well he nailed (certain unspecified things) in advance—both in absolute terms & relative to the impression one gets by following him on twitter.
EDIT: I’m a dum-dum. You mentioned this tweet, like, two tweets down.
There were none:
I feel like either Alexey is trying to gaslight people, or he interprets the terms “terrifying” and “an existential threat” to mean something quite different to what most people do. The latter feels faintly ridiculous. But only faintly. So much of my life has been wasted due to translational friction that my prior for it being the the source of disagreements is quite high.
If it wasn’t Guzey I would have dismissed the whole thing as trolling or gaslighting, and I wouldn’t have covered it beyond one line and a link. He’s definitely very confused somewhere.
I’m not actually familiar with him. Does he have any takes on alignment worth reading? Or on anything more generally?
For what it’s worth, he has shared (confidential) AI predictions with me, and I was impressed by just how well he nailed (certain unspecified things) in advance—both in absolute terms & relative to the impression one gets by following him on twitter.