Now, *why* Turing-completness *wouldn't* be enough?

[sir_pudding]

Originally Posted by vicky_molokh

But then there's ZMC who thinks that souls are required for AIs (if I remember his position correctly), and other less recent protests against the possibility of infomorphs.

I'm pretty sure that's not Ze's position. He's an atheist, AFAIK. He seems to think that the hardware is what's important and that it's not superscience to have AI; just that it's superscience to treat one set of hardware and software as a copy of another set. Or something like that. It's confusing. Like I said, I'm pretty sure he's what my Grandma calls "ornery".

[vicky_molokh]

Originally Posted by sir_pudding

I'm pretty sure that's not Ze's position. He's an atheist, AFAIK. He seems to think that the hardware is what's important and that it's not superscience to have AI; just that it's superscience to treat one set of hardware and software as a copy of another set. Or something like that. It's confusing. Like I said, I'm pretty sure he's what my Grandma calls "ornery".

That's what I said - that he thinks that souls are required for Infomorphs, and consequently Infomorphs are not possible.

[Anthony]

Originally Posted by vicky_molokh

It is an interesting thing to question, but I'm not sure an alternative is even possible.

Doesn't matter for purposes of this argument. If the mind is a finite deterministic algorithm, it can be modeled by a turing machine, and therefore if you hold that turing-completeness is not sufficient for a mind, you must also hold that the mind is not a finite deterministic algorithm.

[sir_pudding]

Originally Posted by vicky_molokh

That's what I said - that he thinks that souls are required for Infomorphs, and consequently Infomorphs are not possible.

He seems to accept that it's possible to copy the memories and instructions for one AI to another set of compatible hardware.

[Jeffr0]

One argument against this goes something like this:

1) Hilbert was overly optimistic about the capabilities of mathematics.

2) Godel showed that mathemetics has counter-intuitive limits. (Ie, there are things that you can know to be true, but that you cannot prove to be true within the context of particular axiomatic systems.

3) Infinite numbers behave strangely-- and the diagonal proof is positively mind blowing when applied to Godel's proof and also to Turing machines.

Therefore...

4) Since math is weird and strangely limited... and since 19th century ideas about the universe are wrong... you cannot make a computer that is truly intelligent.

Yeah... it's a goofy argument, but it is enough to sell a few books. You can sell even more if you can tie some physics into it, but I can't really speak to that part at all.

[This is a summary of about half of one of Roger Penrose's books. He is the discoverer of Penrose tiles.]

[wellspring]

Originally Posted by Jeffr0

One argument against this goes something like this:

1) Hilbert was overly optimistic about the capabilities of mathematics.

2) Godel showed that mathemetics has counter-intuitive limits. (Ie, there are things that you can know to be true, but that you cannot prove to be true within the context of particular axiomatic systems.

3) Infinite numbers behave strangely-- and the diagonal proof is positively mind blowing when applied to Godel's proof and also to Turing machines.

Therefore...

4) Since math is weird and strangely limited... and since 19th century ideas about the universe are wrong... you cannot make a computer that is truly intelligent.

Yeah... it's a goofy argument, but it is enough to sell a few books. You can sell even more if you can tie some physics into it, but I can't really speak to that part at all.

[This is a summary of about half of one of Roger Penrose's books. He is the discoverer of Penrose tiles.]

Was the The Emperor's New Mind? Didn't Penrose go on to argue that the brain is highly sensitive to quantum effects? He essentially went on to say that throughout history, we've picked our most complex machinery as an analogy for the brain, and so pretending that the brain is a computer is just one more analog that may or may not correspond to the underlying process.

He didn't really have any proof, but my 3rd-hand, ten years of hazy recollection later understanding of the argument was that he pointed to quantum calculations as being non-random and non-deterministic. (Even then, I suppose you could simulate a mind using some kind of markov chain to simulate a random distribution without actually reproducing it).

I don't buy the idea, and even if it were true then you'd presumably be able to emulate a brain using a quantum computer, but it's a cool hypothesis.

[Anthony]

Originally Posted by wellspring

He didn't really have any proof, but my 3rd-hand, ten years of hazy recollection later understanding of the argument was that he pointed to quantum calculations as being non-random and non-deterministic.

Well, that's certainly enough classify it as gibberish. QM defines statistical probabilities and every evidence we have is that is that it's random within the computed distribution pattern.

[Jeffr0]

The Chinese room thought experiement is a famous argument.

(Note that... Eliza is a completely stupid program that didn't understand anything... but people poured their hearts out to it and talked to it as if it was real.)

[vicky_molokh]

Originally Posted by Jeffr0

The Chinese room thought experiement is a famous argument.

(Note that... Eliza is a completely stupid program that didn't understand anything... but people poured their hearts out to it and talked to it as if it was real.)

I always saw the Chinese Room as an argument that a human cannot be said to understand anything, ever, because a human who knows a Chinese dialect acts in the same position as the Chinese Room in the thought experiment. This happens to be one of the reason why I think Buddhism is up to something (as in, has the right direction of thought), with its ideas of the falseness/nonexistence of self 'in reality'. In essence, a 'self-aware' mind is an algorithm that tries to describe itself, or make a detailed self-reference.

[Jeffr0]

I am a programmer.

I write algorithms every day.

They are the stupidest, most un-aware things I can imagine.

The only reason people think AI is possible is because of romanticism surrounding the computer. It's the sort of thing people felt in the early eighties-- like the guys who made Tron but had no idea what was going on in computers. This is 19th century romanticism displaced to a new medium.

Replace computer with clockwork and you'd have a better example to argue-- can a clockwork be made to think/feel??