@RokoMijic Kind of irrelevant at this point. I'm probably in that range, yet Claude can already do pretty much everything better than me now. My job over the last year has become just asking Claude to build things. Now it's integrated with GitHub I don't even need to publish it anymore.
@IterIntellectus Sucks I can't download this and repost elsewhere. Noone outside of tpot knows how fast this is moving. How long since you posted the other one?
@littIeramblings We've known this for years why are you any more scared than before. If anything things look slightly better than a few years ago as the models at least don't look obviously malevolent
@neetland@RokoMijic If that were true women would be universally attracted to men with high intelligence as that's the main variable which ensures survival of the human species. What we actually have is genius billionaires getting honey trapped by Chinese spies. Men aren't the problem here.
@RokoMijic There are already open source models that can be run on a laptop which a while back would have been state of the art. I don't see how this technology can be limited in the way this post assumes. If we survive we'll always have access to models a year or so behind the frontier
@RokoMijic A good analogy is if ants somehow created a human in an anthill expecting it's first task would be to kill all the ants rather than just going somewhere else on the whole planet since ASI would be able to colonise the whole galaxy
@RokoMijic I'd go further than that. Even if an AI were not aligned, that doesn't necessarily mean it would make extermination of humans it's mission because the nature of silicon is vastly different to biological life and doesn't depend on earth's surface's environment
@RokoMijic We would have got here a year earlier if people hadn't gone overboard in 2023. It's taking actual warning shots to overcome the priors created by the initial boy crying wolf
@RokoMijic Yeah I've seen professional well educated people posting about ai and nuclear weapons comparisons. The general consensus seems to be the public are starting to understand they could be dangerous, but they still don't understand how dangerous.
@gcolbourn I mean yeah 3 years is a short timeline and we didn't know how quickly things would move. We also couldn't imagine how wreckless the labs would be