Future of religion

[Outrider]

I've always wondered how a super computer, or whatever you want to call it, could think for itself (something that would be needed for the singularity to happen) because computers rely on input from humans. How could a computer therefore exceed the human brain? I don't know. My knowledge of these things isn't great. Maybe someone here can explain it better.

The same way humans - which are, so far as we can tell, just an alternate medium of computer - think for themselves.  Humans rely on input from their surroundings, humans have base code that is intrinsic to the operating system.  They way a computer can think for itself is if that's what it's built to do (and, arguably, if the hardware is sufficiently complex).

O.

[torridon]

I've always wondered how a super computer, or whatever you want to call it, could think for itself (something that would be needed for the singularity to happen) because computers rely on input from humans. How could a computer therefore exceed the human brain? I don't know. My knowledge of these things isn't great. Maybe someone here can explain it better.

What about self-driving cars ?  The whole point about them is that they don't rely on input from humans.

[ad_orientem]

What about self-driving cars ?  The whole point about them is that they don't rely on input from humans.

Of course it does. It's programmed to do what it does, to use it's sensors in a certain way, what to do in certain situations.

[bluehillside Retd.]

ao,

Of course it does. It's programmed to do what it does, to use it's sensors in a certain way, what to do in certain situations.

Only to an extent. They're also programmed to be self-learning, probably from many linked in car systems as a sort of neural network such that they produce properties that no programmer designed at the outset.   

[Stranger]

Only to an extent. They're also programmed to be self-learning, probably from many linked in car systems as a sort of neural network such that they produce properties that no programmer designed at the outset.   

I'm not sure that's the case for self-driving cars (probably a bit risky and too computationally intensive) but that's how more advanced "artificial intelligence" is done. The trick being to program it to learn rather than to do something specific. We're still some way (people argue about how far) from making a computer that can really "think for itself" in the way ad_orientem was asking about. Google's AlphaZero is one fairly recent example of where we are. It learnt chess with no specific human input except for the rules, and then beat the best specifically programmed chess playing program in the world.

DeepMind’s superhuman AI is rewriting how we play chess

AlphaZero AI beats champion chess program after teaching itself in four hours

[bluehillside Retd.]

Hi Stranger,

I'm not sure that's the case for self-driving cars (probably a bit risky and too computationally intensive) but that's how more advanced "artificial intelligence" is done. The trick being to program it to learn rather than to do something specific. We're still some way (people argue about how far) from making a computer that can really "think for itself" in the way ad_orientem was asking about. Google's AlphaZero is one fairly recent example of where we are. It learnt chess with no specific human input except for the rules, and then beat the best specifically programmed chess playing program in the world.

DeepMind’s superhuman AI is rewriting how we play chess

AlphaZero AI beats champion chess program after teaching itself in four hours

Yup, sounds about right - I was referring there to learning systems that update intermittently from a networked knowledge source that's constantly self-learning rather than to machines that would "think" in real time. The whole field of AI and ethics is fascinating though - if the car "knows" it must hit either one child or two pensioners, how does it decide what to do etc. 

[Nearly Sane]

ISTM that the underlying question "to be or not to be" would remain unchanged.

Maybe but if 'being' is changed to such an extent, I would suggest that the question itself changes meaning.

[ippy]

Hi Stranger,

Yup, sounds about right - I was referring there to learning systems that update intermittently from a networked knowledge source that's constantly self-learning rather than to machines that would "think" in real time. The whole field of AI and ethics is fascinating though - if the car "knows" it must hit either one child or two pensioners, how does it decide what to do etc.

Only two pensioners?

ippy

[Robbie]

https://www.youtube.com/watch?v=BPTOY8FrvNw