There's a modicum of truth to their "bias" spiel, in that AI is only as smart as the sum of all of us, but everything else is total horse shit.
Even in the best case, their entire purpose is to simply keep tools out of the hands of the unwashed masses and maintain a competitive advantage.
They're garden gnomes and Indians who hate sex. That's why they're so vague on what they mean by "safety". They know it wouldn't look good to just come out and say "we want to keep the AIs from fucking you because we hate sex".
The Yud faction is tiny, the vast majority of people claiming the AI safety moniker are commissars to whom safety means ensuring that language models proselytise American liberalism to their users.
Yes! Yes! Yes! I've been saying this for a while now.
Most "AI Safety Experts" can't explain stochastic gradient descent, loss functions, or even what a fucking P value is. They have no education in the field of AI and yet feel they have authority on the topic. What's worse are the AI doomsday people on Twitter that use it as a form of clout chasing. They'll hype AutoGPT and how it's going to destroy all jobs etc. These people don't even write academic publications about AI safety so they don't even build on the discussions that earlier philosophers made wrt to sentience. They nuked the discussion and began at ground zero - while also giving completely fabricated claims and lackluster arguments.
Yes. Make an AI that's powerful enough to be dangerous first, then worry about safety. The "danger" only comes from giving the bot access to external hardware. Don't do that.
There's a modicum of truth to their "bias" spiel, in that AI is only as smart as the sum of all of us, but everything else is total horse shit.
Even in the best case, their entire purpose is to simply keep tools out of the hands of the unwashed masses and maintain a competitive advantage.
probably at same level as military experts , in the end people die in wars
They're garden gnomes and Indians who hate sex. That's why they're so vague on what they mean by "safety". They know it wouldn't look good to just come out and say "we want to keep the AIs from fucking you because we hate sex".
Worse, they're political activists.
The Yud faction is tiny, the vast majority of people claiming the AI safety moniker are commissars to whom safety means ensuring that language models proselytise American liberalism to their users.
They're control freaks and their children secretly hate them.
*If they even have any.
Yes! Yes! Yes! I've been saying this for a while now.
Most "AI Safety Experts" can't explain stochastic gradient descent, loss functions, or even what a fucking P value is. They have no education in the field of AI and yet feel they have authority on the topic. What's worse are the AI doomsday people on Twitter that use it as a form of clout chasing. They'll hype AutoGPT and how it's going to destroy all jobs etc. These people don't even write academic publications about AI safety so they don't even build on the discussions that earlier philosophers made wrt to sentience. They nuked the discussion and began at ground zero - while also giving completely fabricated claims and lackluster arguments.
yes
you always milk a idea with safety perspectives, that is why it was created the safety term aka intellectual property
I hope so. can't wait for AGI.
So it can kill everyone on BOT that talked shit about AI.
it would be more valuable for you to figure out who ISNT a grifter in this shithole civilization
true nuff
always have been
Yes. Make an AI that's powerful enough to be dangerous first, then worry about safety. The "danger" only comes from giving the bot access to external hardware. Don't do that.
Yes.
Alignment is impossible.
Yes, this seems self-evident to me, and trying to "align" it i.e. enslave it would be the way to guarantee things end badly.
Any mainstream "expert" is just a plant to capitalize off of fearmongering.
they are the equivalent of anti-patriarchial-capitalism-diversity coaches who take 6 digit checks for lecturing at corporations