[Ghostavo]

Hypothetically an infinite switch-case could match human behavior. 

[General Battuta]

I'm not going to argue that these hypotheticals are likely, I'm just starting from the existence of a strong AI - which as I've mentioned before I consider pretty improbable in the near term.

[Ghostavo]

Since I've just started working on my thesis about a weak AI, I'll have to agree with you there.

But while we're on the subject, what would be the goals of a strong AI?

[MR_T3D]

Since I've just started working on my thesis about a weak AI, I'll have to agree with you there.

But while we're on the subject, what would be the goals of a strong AI?

unless its given some input, it will be just like a baby, methinks.
we could try 'teaching' it, but it will able to chose if it wants to remember/follow teachings, so its actual goals are unpredictable.

but part of me thinks that if it observed the world around it, it'd try to reproduce, thinking that'd give it acceptance with the world. IE create another AI, so it has some'one' to be with.
that sounds pretty sappy, but I would not be surprised if that happened.

[Bobboau]

yeah, except as a digital intelligence it would just have to run the copy command for ten seconds and have a small metropolises of it's self.

[MR_T3D]

yeah, except as a digital intelligence it would just have to run the copy command for ten seconds and have a small metropolises of it's self.

part of me thinks that'll be more like suddenly creating multiple personalties, and/or that the AI needs its  own set of hardware to run on, multiple instances of itself would be extremely unstable.

[Liberator]

Any "true" AI along the lines of being discussed, which to me means Asimovian style AI, is going to be something of an "idiot savant" being able to make human like decisions in the areas that it has information about but terribly weak in areas it doesn't.  Until someone screws up and makes a Daneel type and then we're all screwed.