Quote:Original post by arithma A quick and (I hope) specific question. In what ways you as a game programmer think it's possible that your created AI can realize your existence as its creator. Is it even possible for it to realize your existence. How does that apply to us and God?
For the first half, just look at how you intereact with your parents.
For the second, given that God is only an invention of ours, and we have defined him to be outside our world as non-intereactive... I don't think we get too far. Well, say maybe he does exist but never intereacts with us so we have no evidence. If we then assume he exists, then is our assumed god the same as the real one? I'd say no.
Ohwell, I guess I'm trying to say that (if) an AI were sufficently intelligent (they aren not btw) and it was allowed to interact with its creature; then Sure it could realize his existance. For Us and God though, seeing as how God doesn't talk I don't think the same situation applies.
Personally, I feel that the "creation" of the concept of a "God" spawns from the chicken and egg conumdrum. The question is that whether the egg came first or the chicken. The answer is that we don't know, unless we assumed that at one point, someone created the first egg or chicken. (this is of course if we don't consider evolution and such)
So, the spawning of the "God" idea is primarily due to the human need to find origins, kind of like finding the cause to any given effect. We do it in science and many other fields of research. So, for those that believe in evolution, then humans came from evolution. For those who don't, the only logical solution is that at one point, humans were created by someone or something, which/who must have also created everything else (maybe). And thus, the genesis of the "God" concept.
So, logically, it is possible to create an AI program that eventually, through deduction, realize the possible existence of a "creator."
it's a fascinating question, arithma, and I definitely look forward to a more specific phrasing of it; I think in particular, when AI does advance to a level that parallels human abilities, the masses will start to question precisely what "consciousness" and "intelligence" is; begin to wonder what is really so special about humanity, look at themselves objectively.
And that is when worldviews will crumble away, much like globalization and relativistic thought have begun; in this case, it will be much more personal though.
Having a wealth of information on other cultures and religions has caused a lot of confusion for people in the world today, resulting in both fundamentalism and nihilism; but when lots of people start delving into the question of precisely what their own existence is, I think an even more disturbing ripple will surface in global culture, an internal one.
Then again, people deny evolution and God side by side, so maybe it will amount to nothing at all.
Quote:Original post by arithma I had already put it inside that humans are like machines: deterministic.
I know of no real world machine that is deterministic. Determinism is an ideal regarding the ability to precisely predict the future states of a system given all conceivable information. However, we also know that one cannot observe any real system to obtain complete information (Heisenberg's principle at work). So fundamentally, either the universe is uncertain, or information processes are incomplete. Either way, we can never know that a machine is deterministic (either because it isn't, or our knowledge of its state is always incomplete).
Our models, on the other hand, are often deterministic and designed to explain expected behaviour. That doesn't mean they are truthful statements about the dynamics of the system. This is true of every theory of physics and chemistry. Interestingly, we don't have many fundamental 'laws' of biology, as we have, say, a theory of gravity. However, just because a model is an approximation, doesn't mean it has nothing useful to say about the system being studied.
Artificial Intelligence is about building models; not just to explain what 'might' be happening, but also simply to do useful things (to apply to the automation of systems and processes that previously required human control and understanding to make them work). Most people approaching AI from a scientific perspective don't believe that their models are statements of truth about how human intelligence is created. However, hopefully, creating systems that can behave intelligently leads us closer to understanding the factors that are at work in human and animal intelligence so that we can better understand ourselves and our place in this universe. That doesn't mean that our AI models are anything more than models. It also doesn't mean they aren't informative or useful.
* Godel, Escher, Bach: an Eternal Golden Braid - by Douglas R. Hofstadter * The Mind's I - by Douglas R. Hofstadter and Daniel C. Dennett * The Age of Spiritual Machines - by Ray Kurzweil
Krek, you are asuming the universe IS deterministic, while sience is basically just finding out it very well may not be...
1stly there's Heisenbergs uncertainty principle (wich Timkin explained), wich means that even IF we live in a deterministic universe we can't determine all the info needed.
2ndly, the universe probably isn't deterministic; information gets lost in black holes for example. Wich means we can't reconstruct the course of time even if we had all the info needed.
I wasn't saying intelligence can not be explained (it is being explained in phsycology & biology and especially so together), but they can not be explained in mathematical terms, for mathematics is an absolute sience, unlike psychology. Even sience is now dealing more with estimations and probabilities rather then precision predictions (because of reasons explained above).
And on top of that, we can't even say with certainty that free will exists, who is to say that our brains are not simply chemical factories responding in an absolute manner to all it's chemical inputs (things we eat, chemicals our own body's produce in reaction to whatever etc etc). We can't even know for sure if what we think&feel is a creation of ourself or of our environment.
So we really don't have any basis for a discussion on "real" intelligence, by "real" intelligence I simply meant something capable of unique thought unlike one of these online scripts for example, where you can have a grammatically correct (and at 1st sight quite convinving) conversation with a script (wich clearly is not capable of any thought whatsoever). So even if we can perfect such scripts to the level of passing the turinger test, it is not even near what I consider "real" intelligence. We can be certain a script is not intelligent because we can oversee all the decisions it can ever make (it will never come up with something of it's own).
However this thread had lots of interesting links and pointers and I've read up on a few of them including the Numenta & blue brain projects wich already prove that we CAN create the building blocks of intelligence using current languages (wich proves me wrong).
I told you I was no authority, it's simply an interesting subject... :) These kind of discussions educate us all in a way.
That news bit about mice is really fascinating. It's also good to hear that scientists are positive they already have the required knowledge, they just need more power to do it. That'll happen, sooner or later.
The universe may or may not be deterministic, but that is irrelevant because for all practical purposes universe is infinitely complex and thus we would need an infinitely fast computer to predict all possible outcomes.
But we can reconstruct and also predict the course of time to a certain degree. We know how the universe begun and we know how it will end. We don't know if some us will be alive next year, though.
Human psychology is not a science so lets try not to use it in the same frame as biology, physics and mathematics. It has no influence on defining what "real" intelligence is nor how it operates, particularly because intelligence is not something specific to humans. Otherwise, those scientists at IBM lab could very well ask "mice psychologists" a few tips on modeling their algorithms. A psychologist would however make a great asset in making game AI scripts by defining characters of evil, neutral and good allignment. Off course, coming from a psychologist, it means those definitions would end up being specific to a certain culture of a certain species because as we all know, blood elves have different psyche from underground dwarves so you cannot make any cross-species generalization of (ab)normal behavior.
Another AI Philosopher, Hubert Dreyfus, has written a lot about his impression that AI is (in this timeframe) inadequate to be anything other than a simulation. Nothing that I have read of his thus far has answered your question(s) directly, but if I could take the liberty to abstract the answers, I'd say that Dreyfus feels that CURRENT AI could NOT realize it's creator, nor could it realize God or any relations between itself, it's creator and God.
The important point Dreyfus makes is that he is criticizing current AI, not potential/future AI. In that premise alone, I agree. I don't necessarily agree with his more specific ideas... (the Expert-State, is one I don't agree with)
My question to you (as well as the answer) is, would you be sufficient with a simulation of your questions? I share Dreyfus' thoughts in that I don't think current AI is advanced enough yet. But I am optimistic... I feel it is just a matter of time. Current AI can easily be made to simulate understanding, but doubtful as to if it could realize understanding.
A good point one of the other posters brought up... how do we make that differentiation? How do we determine that WE (as living, human beings) understand anything, and it's not just a "simulation". Perhaps, the biggest factor here is, how we came to "know" and "do". With an AI, whatever it is, whatever it knows, it was all "artificially" put into the system. Even AI's that learn - they don't learn the same way organics do. Learning, to a machine, is NOT a very complicated mode. It is basically copy, paste, modify, copy, paste, modify, etc... until the pasted solution is correct. Either way, some other conscious being had to tell the computer to do this... it wasn't intrinsic. As far as we know, we evolved from the first extremely primitive forms of life that appeared on earth some billions of years ago, and we became able to form our intelligence. There is something (even if it can't be defined yet) that is hard to deny that we possess as living beings, which is different than some a computer we forced into "learning" or "simulating".
I have just started my academic track into the field of Artifical Intelligence, and I look forward to the years to come. In about 4-5 years, I'll hopefully have my doctorates with a heavy concentration in AI.
Alex~
"The truth hurts for a moment. A lie hurts for a long time." - Dean C. P.
The idea behind abiogenesis is that life emerged from a series of natural accidents. Life did not just pop out from vacuum - that's the domain of religion claims. And nobody on this forum was born with an empty brain, we were all born with our BIOS and EPROM which allows us to learn and modify our behavior. We all come with initial programming built into our bodies.
Now lets imagine that in the future we manage to create artificial life forms that can learn and modify themselves. In this case, the "accident" that will create them will be us. While the cause of "accident" is different, it is nevertheless natural in both cases.
So does this make "artificial" life forms somewhat less important because the natural accident that created us is different from the natural cause (our own doing) that will create them? Wouldn't that mean that products of our intelligence have lesser value than products of some random radiation, lightning and meteor impacts (whatever abiogenesis ends up being).
