Related questions
X-risk is pseudoscience and arguably a cult. Loudly insisting you are right does not make it true. Just because you can conceive of something does not mean it will happen. You would think that before trying to put the brakes on something as beneficial as modern AI research, one would require a single shred of actual evidence in support of X-risk claims. Yann LeCun is an accomplished scientist who, even if somewhat narcissistic, is clearly capable of understanding complex and nuanced topics. Donald Trump is the leader of a brainwashed cult of personality and a serious risk for the long term health of the United States, and by extension, the world. I cannot believe that this vote is as close it is.
@Balasar I mean, one side has consistently made wrong predictions about "AI will end the world in 5 years!" and it's not Yann LeCun. If the rationalist longtermist whateverists believe what they're saying, a lot of them should be acting a lot differently.
Personally I'd like to see even dog-level AI before we start sounding the "singularity imminent tomorrow!" alarm. It takes time to make things. We aren't going to stumble from where we are now into tomorrow AI apocalypse just like how we didn't get to the atomic bomb accidentally from 1800s chemistry. ChatGPT isn't quite it when I can gaslight it into thinking 2...O-O-O# is a legal move
@StarkLN The sad part is that since their projections are always a moving target, these people will never admit they are wrong, and continue to dig themselves into the hole until people learn to ignore them. Not very dissimilar from most religions which have assured us for millenia that their Messiah/Apocalypse/etc. is always just around the corner.
@Balasar Really wish there was more basic philosophical literacy in general. I feel like it's very easy to get deathtrapped into this sort of lottery ticket but the EV is maybe possibly it sounds reasonable to say positive? type thinking. Not to say that there aren't any good act utilitarian thinkers, but it's getting a bit... how people can so consistently take a set of questionable naive assumptions and assert "it's just obviously true. Any rational person can see that!"
@Goodgulf you don't get a second chance with AI X-risk, and Yann seems quite likely to consistently and competently sabotage the AI safety efforts of others
This "sabotage" and "trying to get killed" is just too much. LeCun is an intelligent person that obviously wants good things. He just seems to think that 1) transformative AI is far away, and 2) alignment is easy enough when it's needed. But he's a scientist first, and when he sees bad things actually happening, he's in a 100 times better position intellectually than e.g. Trump to do something useful. Yes, under some hard takeoff scenario that is not nearly good enough, but so is not whatever Trump would do.
@Goodgulf Linus Pauling was a Nobel Prize-winning chemist, just like LeCun is a Turing Award winner for AI capabilities research.
Linus Pauling was really, really dumb about a closely related but distinct subject than the one he was an expert in, biochemistry - he believed in consuming massive amounts of vitamin C for your health.
Yann LeCun is really, really dumb about a closely related but distinct subject than the one he's an expert in, AI safety - he believes corporate law could control AGI (among other batshittery).
Both of them persevere because they are not aware of their ignorance in the relevant field.
The difference is that Pauling made expensive pee, and LeCun might get humanity extinct.
@BrunoParga and @Goodgulf you both make really good points
I would still lean away from Yann because his policies could make it much more likely that the take off happens within his own presidency rather than whoever gets elected next