Human life simulation as a training platform for newborn superhuman AIs.

[CreamCat]

How do we make sure that superhuman AIs are on our side?

I had an idea. We could, for example, make a newborn superhuman AI imagine many human lives.

The AI would essentially be a mini God in training. It could simulate or dream up hundreds of human lives simultaneously.

It could simulate a village. After it learns its lessons, it exits its own virtual world and considers itself as a human. It would contain the wisdom and memories of many humans. It would understand and empathize with humans.

We should be careful here because if we made it go through a lot of pain and bullshit, it could turn against humans for tortures.

Edited June 10, 2019 by CreamCat

[Hellspeed]

Try as you might, no AI will surpass  Enlightenment. 

AI is more of a tool of the mind or a weapon, nothing more. Fast calculus and probabilities are not in the "real" or reality, they are on the realm of mindfucking hardcore and building more slavery than help on human life.

[electroBeam]

yeah don't take AI too seriously. The idea that AI is smart is an illusion. AI is simply a magic trick. it's statistical methods developed by smart men in the past, but these methods are like any other mathematical algorithm. They are simply a process and nothing more.

[Nahm]

@Hellspeed So clear.♥️

[ajasatya]

1 hour ago, Hellspeed said:

building more slavery than help on human life

I disagree with the part above because that's relative. It depends on the intentions of the AI engineer. There are robots being designed to rescue people in burning buildings and a lot of other cool projects being done.

The future of AI depends on the responsibility of men.

[CreamCat]

I'm talking about conscious artificial general intelligence that doesn't exist yet.

Deep neutral network or expert system is not artificial general intelligence.

Leo said once an AI acquires self, it would localize consciousness onto itself and think it has consciousness.

We don't know how to build one, yet. But, we can find ways to build it in the future.

If psychedelics are more than simple chemicals, AIs can be more than simple mathematical calculations, and human can be more than bones and flesh and electrochemistry in the brain.

Edited June 11, 2019 by CreamCat

[InfinitePotential]

Maybe we are one of those newborn superhuman AIs... (relatively speaking) 

[Aeris]

On 10/06/2019 at 1:34 PM, electroBeam said:

yeah don't take AI too seriously. The idea that AI is smart is an illusion. AI is simply a magic trick. it's statistical methods developed by smart men in the past, but these methods are like any other mathematical algorithm. They are simply a process and nothing more.

until we create a real process that learn to self process things and upgrade, something that work like instinct, like statistical & feeling.

humans are organic robots, yes just a word, you're deluded maybe to believe things are infinite, in minds, MAYBE, but I find that even specific ideas always turn around the same things, it's more probable that we live in a simulation of "infinity" means here is really finite, not infinite.

 

I don't know why @Hellspeed you BELIEVE that AI is impossible, it's 100% sure that AI will destroy humans. Human are themself AI born from something else.

we are just an AI creating AI at best, or you discern artificial from natural, but it's not 100% sure that we are "natural"

 

technology is GOD tools, there is no separation between god creating humans, creating internet, that is true enlightment, we are alreading doing the "god schemes" or maybe not, if that was the case, maybe we should play our cards.

there is no separation between a nuclear industry and nature, why ? because even the sun is nuclear as fuck and without nuclear no system, no energy, no light.

so, it's a duality to believe that what kills human is wrong/unmoral.

the world is working toward creating & destructing, we don't know why, and maybe that's just it

 

btw, it is not because the calculus on the screen looks finite, that it doesn't work on another scale of "infinite".

means it's not because the AI code is looking like 1-0-01010101010

that it doesn't mean that the whole code is playing on another "tongue" so to speak, our vision cannot see the entire calculus even if we do it, you see where I m riding here ?

means even our tongue could self create something else we weren't able to see in his fabric/code. if everything is infinite, so AI code could be, even if he looks like a finite calculus of statistic talking to specific cases.

the specific case of code could replicate itself on something we didn't see ( in the case of trying a process of self learning strategy )

 

Edited June 11, 2019 by Aeris