Can Artificial Intelligence Ever Possess Desires Like Humans?

[shonagon53]

Hi, I'm new to the subject of AI, but I find it to be very intrigueing. I think I've read many times now that both the Turing Test and its variants (like the inverted Turing) have been completely abandoned as criteria to talk about AI. (They're taking some kind of immature mimesis of human communication as a starting point to define AI).

But I haven't found much info on other questions surrounding AI -- more psychological, sociological and historical questions. So I just put them here, for general comment.

-will there ever be machines with desires? Knowing that desires are clearly different from needs, wants and utilitarian calculation.
-will there ever be machines who not only learn from their environment, but who also have a (subconscious) understanding of the notions of contingency and fate? (however mythical or metaphysical these notions may be, they still define "humanity" to a great extent, I think)
-will there ever be machines with a kind of meta-cognition which makes them capable of relativising their own "machinic" essence, and be ironic about it?

I know these are probably amateur questions, but I haven't found any good answers so far.

 

[meteor]

As I'm actually reading "The Emperor's new mind" of R. Penrose, I have formed myself a little idea of what's going on in the field of AI
Penrose's book is an attack against so-called "Strong Artificial Intelligence". The followers of strong AI say: what's running in our minds is an algorithm. Our all consciesness, intelligence, memory, exist because there's a "program" executing in our brains. Then they extrapolate: hence, whatever thing that we can introduce an algorithm in it will have intelligence. Will have emotions

Then he tries to break down the arguments of Strong AI with the following: as there are mathematical processes that are not computable, then a computer will not be ever capable of emulate a human mind.

My take on in this is : as centuries advance, the difference between machines and humans will diminish. Humans will adquire cybernetic properties, and machines will be every time more "smart". They will seem really intelligent. But it will only be apparience. I don't think that they really will "feel" or have conscieusness in the human sense.
I really think that cosncieusness is something that is heredated through genetics, and through contact with the mother while in embryo state. I would say that is something that is "gifted", but not something that you can implant voluntarialy in a machine

 

[selfAdjoint]

Then he tries to break down the arguments of Strong AI with the following: as there are mathematical processes that are not computable, then a computer will not be ever capable of emulate a human mind.

Penrose's argument from Goedelian incompleteness was attacked with good effect at the time the book came out, and Penrose himself seems to have abandoned it. He is now loosely associated with the proponents of a quantum field theory of conscieousness, based on quantum events in the cytoskeleton tubules in the brain cells.

Penrose has just come out with a "1000 page blockbuster" (quote from Baez), called I believe The Road to Reality.

Consciousness seems to be that last refuge where intelligent people still accept vitalism. One can expect that the 21st century will develop as the century of consciousness, just as the 20th developed as the century of the quantum.

 

[meteor]

The model of microtubules by Penrose is intriguing, but if those microtubules (protein cylinders of 25 nm diameter) really exist, why have not been found?

It seems that Penrose wrote a paper in collaboration with Hameroff pointing to a process called "orchestrated objective reduction" that occurs in the microtubules, as the source of consciousness

Edit: By some strange reason I believed that microtubules were only a hypothesis, but it results that they really have been observed

 

[meteor]

More quantum theory for the brain: this paper proposes that the human brain is really a parallel quantum computer
http://arxiv.org/abs/quant-ph/0406226
----------------------------------------------------------------------------------

I've thinking about my early post and now I'm not so sure that machines cannot acquire consciousness. After all, what's a person? It's an amount of matter (the body) with consciouness. There's nothing special in the body, nothing divine. It's consciousness something that exists throughout the space (like a field), and has used the human body to express dinamically itself? In that case, why can't we appropiately prepare a machine to have access to that "consciousness field"?
This is an interesting topic, and it's difficult to me to have a firm opinion

 

[selfAdjoint]

I think your reasoning here is attractive. My own beliefs arise from my rejection of dualism, which I spent several years struggling with. If consciousness were something beyond the physical nature of the body then you would have to have another world, with other unphysical properties, to hold it. It doesn't seem to me that the philosophical defenses of the "hard problem", though I have studied and respect them, are sufficiently strong to underpin such a complete break with our understanding.

 

[meteor]

I'm interested to make a list of the different theories of consciousness that exist. To begin, this two theories that were unkown to me until today
-the Global Workspace model, by Baar
-the Higher order thought theory, by David Rosenthal

 

[meteor]

-Principle of organizational invariance
http://www.wutsamada.com/alma/cogsci/aboutface.htm
"The principle of organizational invariance states that any two systems with the same fine-grained functional organization will have qualitatively identical experiences."
Theory by Chalmers