Hehe... I got the sorts of responses I expected. 8)

First up, my comment was about the 'illusion of intelligence'. This has nothing to do with the existence of actual intelligence, but merely what someone thinks they see. Since it is in illusion, it must be an aspect of perception.

Again, 'appropriate' is in the eyes of the beholder, so it is only necessary that the beholder see the chane of behaviour as appropriate. I chose the word entity because some people don't like agent. It wasn't intended that this should refer to objects that cannot create internal changes via external stimuli.

Trapper, I think your example is a little far fetched, but I do get your point. I originally worded this statement in terms of agents and actions, but I generalised it to give it as broad a context as possible.

While the statement I made could be applied to the perception of 'intelligence' in computer game agents, it also has as much applicability in the perception (or inference) of intelligence in other human beings.

My final comment for now: I don't consider internal drives like the drive to eat, procreate or stay warm as having anything to do with intelligence. All evidence points toward such behaviours as being hard-wired, or at least encoded dispositions that govern most of our actual behaviours.

Anyway, thanks for the input guys. Feel free to add more... I'll keep reading the responses (but I think I got out of this what I needed).

Cheers,

Timkin

To me, your statment describes a Robot or Zombie rather than an Intelligence.

The distinction being that in addition to being an 'appropriate' change of behaviour, I'd add that it be 'nondeterministic/psuedorandom'

So, a robot that does random actions -which do not get it killed
gives the Illusion of being intelligent more than a robot that does always the same action -no matter how logical or safe it is for its purpose


Why? Well, I think that a 'true' intelligence is so complex that it cannot be understood by ourselves, the limit being that we are examples of such and self-emulating is hard... Thus a lack of understandability gives the feeling of something being intelligent. Random yet workable behaviour is a way to approximate this lack of understanability.

Quote: Original post by Anonymous Poster
The distinction being that in addition to being an 'appropriate' change of behaviour, I'd add that it be 'nondeterministic/psuedorandom'

I think I'd take the position that even a deterministic change of behaviour could still be perceived as intelligent, so long as it was also perceived as appropriate to the stimulus. Only after observing several such occurrences of the same stimulus and response would one be likely to proclaim no intelligence, but merely physics.

Let's change the discussion around a little... what do other people think constitutes the perception of intelligence. I.e., what do you need to see to know you observed an intelligent agent in action?

Cheers,

Timkin

Having the intelligent agent not commit the definition of insanity. For instance, the agent is in a room with multiple doors. And there is a large human sized glass, completely transparent, with something valuable inside. The agent walks toward the valuable item and runs into the glass (not knowing it was there of course). Now it's allowed to try to get to the valuable item once an hour. If the agent keeps running into the glass. It's not intelligent or has very low intelligence. If it at least checks for the glass, then there's there progress. If it tries one of the doors, then there major progress and that'll be a sign of intelligence. For me anyway.

Quote: Original post by Timkin
First up, my comment was about the 'illusion of intelligence'. This has nothing to do with the existence of actual intelligence, but merely what someone thinks they see. Since it is in illusion, it must be an aspect of perception.

Again, 'appropriate' is in the eyes of the beholder, so it is only necessary that the beholder see the chane of behaviour as appropriate. I chose the word entity because some people don't like agent. It wasn't intended that this should refer to objects that cannot create internal changes via external stimuli.

The only suggestion that I would make is to define/add the beholder.

The 'illusion of intelligence' is the perception by an observer, of an appropriate change(s) of behavior of an entity induced by an external cause.

I have been personally focused/interested in emergent behavior, so the original statement rang true (albeit maybe a bit stiff [wink]). If the class is on emergent behavior, might I ask what you are using for a text book? I am always on the look out for a good book on the subject.

Quote:
The 'illusion of intelligence' is the perception of an appropriate change of behaviour of another entity, induced by an external cause, for the betterment or benefit of that entity.

Just a though. It just expounds on the 'appropriate change of behaviour' a bit more.

A football, to a degree, does demonstrate intelligence (if properly modelled). If it is kicked like a football, it ought to fly like a kicked football, not a golf or tennis ball. The question is though, what separates something like a football, to say, a rodent? My thought is that it chooses it's behaviour on it's own, yet that behaviour is still in accordiance with a set of logical rules.*


*Of course, us being human, are not always logical and therefore do not always behave as expected.

Illusions fall apart if they are looked at too closely. There is the element of 'good enough' which needs to be produced to meet the expected audience and stage.
How the performance can be observed usually is carefully manipulated to preserve the illusion.

First it's possible to do a real intelligence on computers. Second change in behavior could be created by internal cause.

As for that statement. It's typical technobable, or better to say scientific speech, I'd recommend to change it into more readable form. Without context it would be hard to agree, or disagree with that few words.

Quote: Original post by Timkin
Hi guys,

I'm presently re-writing a subject on game AI and I wanted to get your reactions to the following statement (so that I can gauge the types of responses I might get from students):

The 'illusion of intelligence' is the perception of an appropriate change of behaviour of another entity, induced by an external cause.


If you choose to respond, could include whether you agree or disagree with this statement and given your position, briefly why you think that way. What I'm specifically interested in is what you think this statement means and implies.

Cheers,

Timkin

What if the change of behaviour was inappropriate but only with respect to your own limited boundaries - meaning the change was in fact 'intelligent', whatever that is, but beyond your comprehension. To a third observer, then who is seemingly intelligent, the observer or the entity? For a weak example: chess gambits by grandmasters vs novices.

The questions are not meant to be answered but rather, to highlight what i feel are weaknesses in such a summation by poking at the extreme edges.

Quote: Original post by Timkin

Quote: Original post by Anonymous Poster
The distinction being that in addition to being an 'appropriate' change of behaviour, I'd add that it be 'nondeterministic/psuedorandom'

Let's change the discussion around a little... what do other people think constitutes the perception of intelligence. I.e., what do you need to see to know you observed an intelligent agent in action?

Cheers,

Timkin

Consistent but varied seemingly autonomous behaviour with respect to stimuli and the convergance of observed outcomes to comprehensibly acceptable values over a period of time?

I like to think I meet this criteria myself. Though I perceive myself to be intelligent, I suspect that I cannot perceive the intelligent behaviour of an entity sufficiently more intelligent. At least not readily, with the amount of observations and length of time required increasing with difference in intelligence.