excreationist

Bill:
But then our brains use this information to make inferences so we learn to have "common-sense". So we can apply patterns we learn in one problem domain to another without being told to do that by another person.
The alternative is an "expert system" which has explicit rules - e.g. Doug Lenat's <a href="http://www.cyc.com/overview.html" target="_blank">Cyc</a>.
It probably still has problems with common-sense though. I guess the problem is that inferences aren't totally logical since you are making a decision based on incomplete information.

Anyway, my point is that the way that chess pieces should be handled is defined in Deep Blue's program. But humans aren't preprogrammed with knowledge about a lot of things - e.g. say you gave them this problem - "orange and red makes orangey-red... what does purple and pink make?" They wouldn't have been taught how to do that explicitly before. If they don't know, you could say "as another example, brown and red makes browny-red". So basically we can work it out. Though we learn the advanced major patterns from others (like English, maths, etc), we fill in most of the gaps ourselves.
That's what I mean by the chess computer being explicitly programmed - it can't just learn the rules and play some games and work out the strategies for itself - it is taught the strategies. People usually benefit from learning strategies from others, but they still refine them themselves.

It is possible that chess computers could learn from their mistakes, but that would probably involve a huge amount of computation for it to analyse itself... so it would probably be worse at chess than a programmed chess computer with the same computing power.

Just about the future of AI research:

<a href="http://www.cyberlife-research.com/about/brainintro.htm" target="_blank">Lucy project info</a> (created by the "Creatures" games programmer and author of "Creation: Life and how to Make it", <a href="http://www.cyberlife-research.com/people/steve/index.htm" target="_blank">Steve Grand</a>)
<a href="http://www.salon.com/books/int/2002/01/02/grand/print.html" target="_blank">Steve Grand interview</a>

Basically I agree with him and I expect there will be AI's with human-level intelligence in the coming decades. (maybe in 50 years)

excreationist

About Deep Blue again:
I suspected that it didn't learn from its mistakes (like we do). Well here is some information about that:
<a href="http://www.defencejournal.com/globe/sept99/kasparov.htm" target="_blank">Globe Magazine Sept 99</a>:
So it is just that it can't learn from its mistakes for itself. That's what babies do when they're learning to walk, etc - they hit a wall or the floor and avoid that behaviour and keep modifying their own behavior until they succeed. So the problem solving strategy evolves within the system to suit the problem domain.

Kip

Deep Blue only plays chess because that is Deep Blue is programmed to do. The conclusion "a machine can never do this or that" does not follow from this premise "this given machine can not do this or that". Your DNA is far more sophisticated (and inefficient), and accomplishes far more than Deep Blue's programming. Deep Blue's computational power is also several orders of magnitude less than that of the human brain (as of now). There are many programs that may learn from their own mistakes. I have no difficulty conceiving of a robot that is programmed to recognize that:

Although induction and abstration is admittedly a difficult program for the machines of today, such behavior only demands sufficiently comprehensive deductive methods, such as genetic algorithms, to hypothesize rules about the world (such as gravity) and test their validity.

[ March 27, 2002: Message edited by: Kip ]</p>

excreationist

I didn't say that in what you quoted. I was just clarifying what I said about Deep Blue compared to animals and people.

Anyway, in this thread I later said this:

Kip

Your comments implied a (popular) distinction between the capabilities of men and machines, if that was not your claim, then my reply was inappropriate.

parkdalian

Hello to all, I'm posting for the first time.

I think that the whole "free will" concept to be nebulous and in more ways than one at that.

Perhaps the best way to define it is in terms of its intended use as it was conceived by philosophers. In order to hold us responsible for our actions, we use the term "free will" to describe the act of deciding something or choosing a course of action without an outside influence. The word "free" is added for the theological slant, i.e. we are absolutely free to choose evil therefore any subsequent punishment is "deserved", implying a vengeful and righteous judgement as opposed to a more pragmatic, deterrent-driven justice.

I believe we are responsible for our actions in the sense that we are aware of the consequences of our misdeeds and we accept punishment as the way the human world works to prevent evil. We want to work out a way to get along with each other and we do it with a system of rewards and punishment and not simply because an invisible god said so.

I also believe there is an inner moral component to that deterrent -- a self deterrent -- the sense of guilt we feel for evil-doing. Even that inner component can be broken down into other elements, from the threat of friends' and families' anger or hatred to the more purely instinctual and altruistic forms of genuine remorse for hurting loved ones.

We do use some system of reward and punishment no matter what culture or society. Whether or not people in general understand how and why it works are interesting questions.

It has been pointed out by many others that the deterministic/random debate re: free will is irrelevant. Who cares if there is a completely random dice roll in my head or if I'm an automaton. So what (as far as the discussion of free will is concerned) if an act of evil originates within me if only because of some malicious quirk I was unlucky enough to possess. The quirk may have originated soley within me but I was not responsible for my own existence in the first place.

Either there was a logical, coherent course of events that led to me having the malicious quirk or one day something went "poof" and I became inherently evil in the act.

Try to define free will in borderline cases such as in the mentally impaired adults or in children of varying ages. Where does free will begin and end in our own development or in the differing circumstances we encounter? If "free" will is really completely free, does that imply we may act or decide upon all strictly yes or no propositions 50 percent of the time, either way?

parkdalian

DRFseven

A example I have put before people: the case of an infant born to incompetent, drug-addicted parents who, alternately, viciously beat and neglect him. Though there is some disagreement between respondents on the culpability of the parents, few argue that the child has not been victimized. Then the child, predictably, begins to develop asocial behaviors as a defense mechanism against his tortuous life, and at the age of eight, waits outside the doors of pubic restrooms for unaccomanied four and five-year-olds to enter, at which point he often follows them inside and brutally beats them. The victim is now the predator, but at what point did he cease being the victim and become the predator? At six, when he began pushing unsuspecting children off the slide, at four, when he was found attempting to choke an infant, or at eighteen months when he gouged at the eyes of a neighbor's puppy with a knife? Or is he simultaneously both victim and predator? Certainly, he has chosen his behavior in that no one forced him to commit these acts, but, then, what does chosen mean?

owleye

Bill

Just found this assertion of yours, ...
"However, both humans and computer are essentially algorithmic engines. You feed in certain stimulii and you get certain expected responses"

supported by:
"(just ask any advertising executive - this is the nature of what the advertising industry does)."

There are undoubtedly several difficulties with what I can only think is the drawing of an analogy between computation (calculation) and human cognition, though it might not be, but the most striking is your contention that because we are influenced by advertising, that our responses to them are the result of a determinate algorithmic process.

What makes you think this is the case? Isn't marketing rather a statistical science, if in fact it is a science? I don't get any sense in which there is any causal determination needed, though they would like it, of course. As Whitehead observed, we are not captured by stimuli in the way kittens are, by a ball of yarn, though of course, we can be, some more than others, I'd imagine.

I would like to add my own view on this, with respect to the influence of advertizing. One thing that differentiates our level of maturity is the degree to which marketing affects the decisions we make. Indeed, the bombardment to keep us young and dumb, is quite delibertate. And it is for this very reason that I think advertizing to children should be banned. In any case, I would hope this is not all you would rely on to make your case.

owleye

DRFseven

The difference is only that we have access to more alternatives than the kitten does. It's still a trip swich that activates our response, even if the response route is much shorter in the kitten. Ours is more complicated because it involves a lot of memory links that the kitten does not have, which makes the kitten's response much more predictable merely by virtue of there being so few alternatives. Still, we are bound to consider or not consider certain possible scenarios and that consideration is set in motion by our physiological state at the moment the stimulus is delivered, just as the kitten's response is determined by its physiological state at any moment.

DRFseven

I agree, but this is not the only factor. Personal experience involving advertising (learning due to experience of direct consequences or absorption of attitudes) is a factor and so are inborn biological proclivities.

I certainly agree with you here. I remember when the FCC stepped in to ban the use in children's programming of commercials that instructed children to "ask your Mom" to buy certain brands of cereals. Children are defenseless against this type of control, as they are in most other types of control, but in the case of advertising, it is often used to enrich the advertisers at the expense of the welfare of the children and/or their families.