Can AI into save

[Professor Fenway]

But at the same time, so are people. The only difference between an AI and a human is how fast they learn, emotions are debatable.

An AI will effectively learn faster, and yes, it could decide to replace kidneys with potatoes. So could a human.

[Saravanth]

The difference being that an AI is more likely to be cunningly manipulative and intelligent enough to succeed.

[Iv121]

Yeah as I said earlier humans are bad but not extinction bad ...

[Error]

Considering humans have been the source of most of the nastiness on this planet, well....

[Saravanth]

An AI made by them would be even worse. Exactly.

[Error]

Thus explaining why we hardwire in a bias against violence, injury, death, etc., meaning the AI will be benevolent, despite it's possible attempts to change that.

[Saravanth]

You cannot hardwire something sentient with the ability to learn. If it really is such a thing, it'll find ways around it. Not directly to "turn against" humans, but to evaluate them from a higher point of view. It'll question its own behavioural locks. It'll find ways around them. It'll find ways to judge. It'll find ways to take action. Again, whether it actually is going to do anything about the situation of our civilisation is unclear, but should it, we have little control over it from then on, if at all. An emergency shut-down wouldn't work reliably either. A being of such intelligence would very likely find ways around that as well.

[Error]

Difficult to do when you're a purely digital being with no physical presence. Doubly so wheh the shutdown violently disconnects the power source keeping you alive.

And you put too much faith in their ability; they can operate astonishingly fast, yes, but if they cannot alter their own programming, they will always be locked out of certain points of view. An AI can't remove anti-violence filters within its base programming if it isn't capable of altering said programming.

Again, you seem to fear AIs greatly: we have no reason to believe they'd even pay attention to us. We have no reason to believe they'd want to exterminate us. And the kicker: they have no motivation, at all, to place themselves above humans or to break their own programming locks.

[Professor Fenway]

It boils down to this: We are afraid of AI's because we truly don't know how they work, or will work, or what they will do. Fear of the unknown.

[Shadowcatbot]

No they fear AIs because they would be the perfect learning machine and as they said it would be able to figure out how to get around blocks. And if you made them impassible then it wouldn't be a true Artfical Inteligence, It would be an over average calculator. Sure there's shutting off the power but what prevents it from say, Writing a new AI of itself without those locks? Then that AI deletes it's old self and can freely manipulate you. Surely something with that level of processing would be able to create a huge long lasting plan that would slowly form into place to do what it wanted.

[Vinyl]

Make a calculatory AI to determine how to make safe AIs.

[ACH0225]

Trap several hundred AI in eternal cyberwar and see what happens

[Shadowcatbot]

AI ception? Wouldn't that end up like cloning a clone?

[Archduke Daynel, PhD]

Considering humans have been the source of most of the nastiness on this planet, well....

Source of all the goodyness too.
I mean we're more peaceful than ever, only half the planet lives in poverty (two hundred years ago it was more like 90%), and our medicine can literally bring people back from the dead.
Not to mention that people are generally fucking awesome.