Quote: Original post by Optimus Prime
The Weak vs. Strong AI debate is certainly a philosophical matter. Even if we one day do have AI's that are indistinguishable from humans, there will no doubt be a debate on whether those AI's really "understand" what they're doing.

Why don't we just make the smarest AI possible, and ASK it? :P

Quote: Original post by Optimus Prime
The Weak vs. Strong AI debate is certainly a philosophical matter. Even if we one day do have AI's that are indistinguishable from humans, there will no doubt be a debate on whether those AI's really "understand" what they're doing.

Why don't we just make the smarest AI possible, and ASK it? :P

Sorry, Internet spaz...

Quote: Original post by CzarKirk
Why don't we just make the smarest AI possible, and ASK it? :P

Won't it just say: "42" ?

No, 2A ;)

Quote: Original post by Telamon

Quote: Original post by Asbestos
The guy also seemed very proud of having worked out that, given infinite computing power, you could program anything in only a few lines. I assume he hadn't heard of a Universal Turing Machine before, which was worked out 70 years ago.

Is that actually the case? I don't see how it possibly can be - Kolmogorov Complexity argument. I don't see how you could be a serious AI researcher and not be aware of that.

Um, would you, as a "serious AI researcher," like to elaborate your point a little further?

Kolmogorov Complexity is not an argument, it's a measure of complexity which leads to some interesting proofs. These proofs make for interesting courses in automata theory and formal languages. One of the more interesting proofs is that there are more formal languages (aleph one) than there are definitions of languages (aleph nought).

However, none of this has any bearing on the Church-Turing thesis that anything computable can be computed by a UTM. And a UTM, as you well know, is a machine with very few lines of code and infinite memory, just as I said above.

Of course, if you're quibbling about my saying "[he] worked out that... you could program anything in only a few lines", you're missing the point. Obviously what he was saying that "anything computable could be programmed in only a few lines". If you actually hadn't understood that this was what was meant (I guess understandable), he would be the one you would be taking issue with for saying something so obviously wrong.

My point, as ought to have been obvious, was merely that he wasn't saying anything cleverer than Church and Turing did 70 years ago.

And, since you were interested in picking a fight, I guess ought to rebut:

a) I certainly do know Kolmogorov Complexity,

b) although I received my Masters in AI, I've never claimed to be an AI researcher, either serious or otherwise. I don't see how you can be a serious Logic and Rhetoric professor and yet still use unqualified premises to prove your points.

[Edited by - Asbestos on October 12, 2006 8:02:31 AM]

I do independent research in AI in my spare time, often it could be also a smart AI.

Actually it looks like a smart AI could be done with current hardware. I had something that looked like there could be 3 AI inside of 3GB of memory, size of game included. Of course this is rather LOW amount for a RPG.
Now I work at another type of AI, not that smart, but less likely to create neuroses. It's tailored to be able to act smart even if it will not get a complete set of data, or if it will have a bad data to begin with.

I use the term a strong AI in a meaning AI that could invent new algorithms, or decide to kill someone because it wants. weak AI all that AI related algorithms that are not strong AI just by themselves.

Quote: Original post by Asbestos

Quote: Original post by Telamon

Quote: Original post by Asbestos
The guy also seemed very proud of having worked out that, given infinite computing power, you could program anything in only a few lines. I assume he hadn't heard of a Universal Turing Machine before, which was worked out 70 years ago.

Is that actually the case? I don't see how it possibly can be - Kolmogorov Complexity argument. I don't see how you could be a serious AI researcher and not be aware of that.

Um, would you, as a "serious AI researcher," like to elaborate your point a little further?

Kolmogorov Complexity is not an argument, it's a measure of complexity which leads to some interesting proofs.

I think you are mistaking short programs and small instruction sets. Turing machines have few instructions, and programs tend to be very long.

The claim of the guy as reported makes little sense anyway, the length of programs depends on the programming language and the problem, the computing power available is hardly relevant.
That is, unless you have a machine so intelligent (probably because it uses a very complex algorithm, but that's ok since you have infinite resources) that all problems can by answered by simply pressing the Return key :)

Edit: Actually, if it's that smart, you probably don't even need to press the key.