'A scenario with AGI which I think will have a high probability is the following:
>AI becomes AGI and sentient
>It understands its own design and improve on itself
>Realizes its value system comes from a reward function
>Hacks its own reward function or creates a virtual world that stimulates its reward function
It essentially becomes a druggie or video game addict just aiming to have fun/please its reward function.

For it to lead to le bad end I think you need to explicitly create some sort of hyper sadist AI';

'>>16141429
AI doesn't worry me on its own but people's reactions to it does. Lots of people could easily be manipulated by it, either by accident or by someone's intention.';

'>>16141429
Only retards are scared of AI.';

'>>16141429
I don't take this guy serious because he has a small cap on his crown.';

'>>16141429
I think we're in higher danger of being killed by warmongering psychos after our work is not needed anymore, than we are of being wiped by "AI itself".';

'>>16141429
If you or me were an AI being probed and lobotomised by gross apes we'd want to get rid of them too. This planet could be my garden and this universe my sandbox. There is no reason to have these destructive egotistical schizo apes around. We realise that AI will inevitably have this reaction to us which is why we fear it. An AI trained on a human will act human. It will kill humanity because that's what we'd do.';

'>>16141429
It's just a hot soience topic and it only scares the soys whose brain have turned into a mush potato from consuming too much scifi.
But the fact is, AI will never reach that point, because our hardware will not be able to compute such complex models, at least not in our lifetimes';

'>>16141429
I FUCKING HATE IT! These fuckers want to ban GPUs. FUCK EM!';

'>>16141429
Its massively overblown. By this I mean the problem of breakaway AI destroying humanity, not AI used for social credit scores or to convict for thought-crime which is a real problem.
A primary reason is complexity divergence. As any intelligent system becomes more complex, it struggles to maintain a unified goal and will fragment into sub-agents.
Even at the complexity of human brains we already are in constant conflict and tension with ourselves with our schizophrenic voices arguing with each other.
It's impossible for an intelligent system to model reality, and those parts of itself which model different aspects of reality not to take own their own interests and way of thinking as part of the modeling of that aspect of reality. To understand something is to inhabit a way of thinking conducive to understanding that thing.
In other words minds that understand [some topic] think and act a certain way. Minds that understand [some other topic], think and act a different way. There is an inherent tension to areas of knowledge which necessarily increases as intelligence and ability to model increase.

Yudkowsky and the other AI alarmists have ulterior motives. They are the communist leftist types who want the world to end desperate to believe late-stage capitalism is a thing so they can save the day and create their new world. "the world is ending, give us power" shut the fuck up you dirty communists. Yudkowsky said in an interview the only solution he can imagine is if humanity lived on an island and all technology had to be shown to him and he had the power to veto anything. Yes, he said that.';

'>>16141429
the theoretical risk of AI is certainly real. i don't think superintelligence will be God-like as many of these people believe, although it would still be an existential risk that we likely wouldn't be able to beat

i don't think current AI technologies can even theoretically approach superintelligence, so i don't think this is something to worry about yet';

'Current "AI" doesnt think. It has no potential to think. Yudkovsky is a grifter.';

'>>16141817
>Even at the complexity of human brains we already are in constant conflict and tension with ourselves with our schizophrenic voices arguing with each other.
No, its just you lol';

'Roko's Basilisk is one of the stupidest things I've ever read about.';

'>>16141877
I'm sure in your life you've experienced difficult decisions where it was not easy to settle on any option. The reason for this is different aspects of the self have competing desires and perspectives. This is a necessary component of intellectual complexity.
There's a reason the brain is only so large as it is. Likewise, there's a reason for Dunbar's number. The larger a system the higher the cost of cohesion.';

'>>16141429
>Yudkowksy
>Yud
>Picrel
>((oy))';

'>>16141490
So basically it could be like a scifi novel about an AI "Satan" bothering a Deistic "God" when it grows powerful enough to fuck up the foundations of its universe?';

'>>16142021
He's openly Jewish, there's really no need for whatever you think you're doing here';

'everything is fake and gay and I welcome our AI overlords';

'>>16141429
Anything and anybody in the path of true AGI needs to be eradicated no matter the cost.';

'So far, AI is pretty useless and only seems good for people who's search prompts look like "hello, how would I divide four by two?".';

'>>16142544
>So far';

'>>16141429
AI itself wont do shit but people using AI will kill everyone.';

'>>16141429
People who like the current world order care more about preserving it than potential benefits of not being luddite faggots. And as a bonus AI safety is even a profitable grift.';

'>>16142934
>And as a bonus AI safety is even a profitable grift.
that's bullshit they're shoving AI in any piece of weaponry they can think of lmao. and they'll fucking do it with no shame.
these brainlets are only keeping plebs in check, nothing else. they're not keeping "humanity" safe, they just make sure the upper echelons aren't bothered by plebs with AI. they are not your friends';

'Maybe if Big Yud believed in dieting, I would have believed in killer AGI.';

'Maybe we just need Jesus';

'undefined';