How to make an AI

[Sam Spade]

Peregrine

Yes, I have seen the matrix, and the sequel, but not the third. As fiction, they make a rather good story: as science, I question the premises made. You can also make the exact argument, and point to "I, Robot" instead. Equally, you can point to Spider-Man 2, where the inhibitor chip overloaded, and the AI of Octavius' arms took over. Who really wants to hear a story of AI going right? It just does not make as good a fiction. Off the top of my head, I can only recall one story of strong AI that does not follow this pattern; Orson Scott Card's Speaker for the Dead.

Golden age sci-fi writers tended to be more benevolent towards AI. The Asimov book "Bicentennial Man" was based on is a fairly benign AI. Also, Robert Heinlein's "The Moon is a Harsh Mistress" features an AI that is mischievious, but not evil at all. Also one of the better conceptions of how an AI would perceive events differently.

[Sotek]

Peregrine

I am actually beginning to believe our fear of a strong AI revolt may be a self-fulfilling fear: someone creates an AI entity, it is censored and inhibited harshly for fear it may revolt, thereby giving it more incentive to revolt than any it might inherently possess.

If I succeed, we don't need to worry about that. ^_-

Quote:

Somewhat perversely, I think an AI revolting would tend to indicate it being strong.

Quite. As would AI protests for equal rights.

[Asharu]

Sotek

Peregrine

I am actually beginning to believe our fear of a strong AI revolt may be a self-fulfilling fear: someone creates an AI entity, it is censored and inhibited harshly for fear it may revolt, thereby giving it more incentive to revolt than any it might inherently possess.

If I succeed, we don't need to worry about that. ^_-

Question: If an AI is created in a standalone box, what have any of us ever need to worry about?

[Sotek]

Asharu

Question: If an AI is created in a standalone box, what have any of us ever need to worry about?

The fact that sooner or later someone would let it out of the box, as it were?

But yeah, paranoia is ... silly, sometimes.

[Asharu]

Sotek

Asharu

Question: If an AI is created in a standalone box, what have any of us ever need to worry about?

The fact that sooner or later someone would let it out of the box, as it were?

But yeah, paranoia is ... silly, sometimes.

Are they worried about the AI taking over the internet, or something?

Make AI's use really lame internet browsers like we have to. Problem solved.

Seriously, the world is not the Renraku arcology, that we have to fear some Deus coming along and pwning our s**t.

[Sotek]

AI are going to be able to operate in Reality sooner or later. But yeah, paranoia is silly.

[Peregrine]

Sam Spade

Golden age sci-fi writers tended to be more benevolent towards AI. The Asimov book "Bicentennial Man" was based on is a fairly benign AI. Also, Robert Heinlein's "The Moon is a Harsh Mistress" features an AI that is mischievious, but not evil at all. Also one of the better conceptions of how an AI would perceive events differently.

Hence, my "off the top of my head"; it has been some time since I read Asimov or Heinlein.

Sotek

If I succeed, we don't need to worry about that. ^_-

Why not? Do you plan on having backdoors to the AI, in a manner of speaking, or some flawlessly coded sentience that would not or could not revolt?

Asharu

Question: If an AI is created in a standalone box, what have any of us ever need to worry about?

Nothing physical. There are other ways of revolting, though: the petulance I conceived of earlier for example, or the deliberate returning of false data. But again, what of the AI's rights? Would not creating an AI in a standalone system be the equivalent of solitary confinement? For the crimes of paranoid authors' writings?

Sotek

But yeah, paranoia is silly.

Most of the time, yes, it is very silly. However, paranoia is not always a baseless conspiracy theory.

[Peregrine]

EDIT: Blasted repeat posting.

[Sam Spade]

Peregrine

Nothing physical. There are other ways of revolting, though: the petulance I conceived of earlier for example, or the deliberate returning of false data. But again, what of the AI's rights? Would not creating an AI in a standalone system be the equivalent of solitary confinement? For the crimes of paranoid authors' writings?

It'd only be the equivalent of solitary confinement if, like a human intelligence, an AI somehow values freedom and mobility and desires contact with other entities. The reason solitary is so harsh is that most people need interaction at some level. There's no reason to presume initially that an AI has the same psychology as humans. Unless of course the method we choose is the exact replication method discussed above, in which case we would have an artificial human.

[Sotek]

Peregrine

Sotek

If I succeed, we don't need to worry about that. ^_-

Why not? Do you plan on having backdoors to the AI, in a manner of speaking, or some flawlessly coded sentience that would not or could not revolt?

That = Installing safeguards against AI revolt thus causing AI revolt.

Also I figure that empathy can't be that hard if the rest of it is doable.

Quote:

Most of the time, yes, it is very silly. However, paranoia is not always a baseless conspiracy theory.

I like what some man said about this, which is "Do you worry about your children taking over?"