My guess is that by unpredictable he means doing the unexpected, yet still moving to acheive the goal, while random would be just doing any strange whacked out thing which may or may not lead towards the goal.

Ok, I think that this thread of what random "is" is getting a little off topic. All I really ment by saying that the AI program should have a random aspect to it is if you were to sit down and talk to it (assuming that it could respond in the same way that we could), if it did not have a random aspect to it, then it would just follow your topics, and wait for you to pause before it put in something on YOUR TOPIC; on the other hand, if the AI program did have a random aspect to it, then it would be able to bring up topics that you would be able to respond to, making it much more intersting, and more acceptable.

My email is ruai@comcast.net

that could well be achieved without pseudo randomness. Perhaps based on curcumstances as just one example, if you are wearing a baseball cap it might start to talk about baseball. If you are naked it might talk about sex, etc.
Take this further and take in thousands of factors about its current environment and it would seem random, but would actually be more interesting. .sen

zzzomed,

What you are suggesting just doesn''t add up. There is no evidence that there is anything ''random'' going on inside the human brain/mind that influences topics of conversation, or any other decision we make. Decisions about what questions to ask during a conversation are usually based on things we want to know, or in some cases, patterns of polite discourse: like asking someone if they had a good day, or commenting on the weather.

I don''t believe it is reasonable to look at current natural language generation programs - which generally exist in isolation from the world and have no motivations generated from ''living'' in the world - and state that they cannot start their own topic of conversation simply because they don''t have a random number generator inside.

Regards,

Timkin

I''m going to stick to Mark Twain''s infinite wisdom of, "better to keep your mouth shut and appear stupid, then to open it and remove all doubt", when it comes to this thread It''s very interesting, but I really don''t have the background to offer any useful insight.

But I thought I''d chime in on something Questionmark said. He quoted Einstein''s famous comment, "God does not play dice with the universe". Well, Niels Bohr, Einstein''s nemesis said something equally amusing. In reference to Einstein''s comment, Bohr said, "Stop telling God what to do!"

And I think that Niels Bohr''s Quantum theories have withstood the test of time better than Einstein....indeed Einstein had to come up with a hypothesis to counter Quantum Theory that virtually destroyed his own theory of relativity in the process. The famous EPR argument which in turn spawned Bell''s Theorem (and was later proved in a laboratory to be correct so I guess it''s Bell''s law now) which shows that "something" goes faster than the speed of light.

quote: Original post by Timkin
zzzomed,

What you are suggesting just doesn't add up. There is no evidence that there is anything 'random' going on inside the human brain/mind that influences topics of conversation, or any other decision we make. Decisions about what questions to ask during a conversation are usually based on things we want to know, or in some cases, patterns of polite discourse: like asking someone if they had a good day, or commenting on the weather .

I don't believe it is reasonable to look at current natural language generation programs - which generally exist in isolation from the world and have no motivations generated from 'living' in the world - and state that they cannot start their own topic of conversation simply because they don't have a random number generator inside.

Ok, I think that what I am seeing as random is what you are seeing as a generalization. You mention speech; we don't say the same word exactly the same time every time we say it (If you don't understand, reread it). (I'm using this as an example, don't go off on a tangent about the human brain and AI ) Our brain sends messages as to how much our vocal cords should vibrate, and how our mouth should move. All of which are generalizations, our brains don't pin-point the exact cordinate to which our cells on our lips should move. As for asking if someone had a good day, or asking about the weather, which one doe we ask first? What determins the first question? It is not alwaies total logical, we do have a random or "generalization" factor to how our brains work.

As I posted before, if the True AI program did not have a random factor, we would not be able to have as good of a conversation to it as we would another human. I have a feeling that someone reading my posts understands me somewhat more than others. I seem to be having a hard time convaying (or simplifying) my message to the other people. I will try my best, but it would help if you told me exactly what you don't understand rather than just:

quote: Original post by Timkin
zzzomed,

What you are suggesting just doesn't add up.

My email is ruai@comcast.net

[edited by - zzzomed on April 27, 2002 11:49:43 AM]

[edited by - zzzomed on April 27, 2002 11:50:15 AM]

quote: Original post by zzzomed
I seem to be having a hard time convaying (or simplifying) my message to the other people. I will try my best, but it would help if you told me exactly what you don''t understand rather than just:

zzzomed, the problem here is one of terminology. Where people apply terminology incorrectly the open the way for misunderstanding. The word random has a clear and distinct meaning that does not mean what you think it means. I''ll get to this later... You were also incorrect to suggest that I am talking about generalisation. I am not.

Furthermore, your last post was insulting by suggesting that we needed your theories dumbed down before we could understand them. I have no time for people that wish to insult me, so if you continue to do so, I simply wont discuss your ideas with you.

As to your points about vocalisation of sounds and apparent randomness in the outputs... this again has nothing to do with randomness in the brain. Muscle excitation is handled by the cerebelum and the spinal cord. The cerebelum emits very specific synaptic patterns (which have been observed in many experiments) to trigger an electrical signal... which is routed by the spinal cord to the particular nerve bundle associated with a muscle. There is likely to be an element of unpredictability in the output of this signal due to fluctuations in the physiology of the person at that particular time, but a) the signal to noise ratio is high; and, b) it is not random, just unpredictable. Finally, the muscle reacts to the electrical stimulus, certainly with some element of unpredictability, again due to physiological fluctuations... but again not in a random manner. The chemical processes going on are quite deterministic. It''s simply that there is uncertainty in the initial states of the system that makes predictability of the final state difficult. This is true for every stage of the excitation process, from the cerebelum to the muscle. Suggesting that there is randomness in the brain that causes the different outputs in the vocal cords (and other physiological mechanisms) is not supported by evidence.

Let''s get something clear: unpredictable does NOT mean random .

Unpredictable means that our state of knowledge is insufficient to determine which, of a number of possible outcome states, a system will be in. It is a deficiency of knowledge. If we had perfect knowledge, we could determine with 100% certainty the output state of the system.

Randomness on the other hand means that even if we had perfect knowledge, we could not determine the final state of the system, as it can be any one of the possible states, with no causal relationship determining that state !

People often confuse randomness and unpredictability because they think that no causal relationship relates to knowledge about the existance of the causal rule. It doesn''t. I means that one does not exist.

Now, in light of these (widely accepted) definitions, would you care to rephrase you opinions so that we can relate to them in a meaningful manner?

Regards,

Timkin

quote: Original post by Timkin

Furthermore, your last post was insulting by suggesting that we needed your theories dumbed down before we could understand them. I have no time for people that wish to insult me, so if you continue to do so, I simply wont discuss your ideas with you.
Timkin

I'm sorry, I should have thought out what I ment to say when I typed put the post more. I did not intend for it to come across to be insulting.

As for the rest of your post: Granted, the AI program will go through a different process in which to come to the same result. As for speaking, I am not trying to say how it works in the body, I am mearly saying how it would be preceaved from the prospective of the mechine which would normally speak one word the same way every time. If the mechine does this, then the human will find the conversation either dry or boaring (seemingly for no reason at all).

If we take away how the human dody physically works, then we will better understand how the AI will come to the same result in forming its words, or anything else that it has to do.

P.S. I am extreemly for implying that I would have to "dumbed down" what I am trying to convay for everyone to understand it. That is not at all how I ment for it to come across.

My email is ruai@comcast.net

[edited by - zzzomed on April 27, 2002 11:09:13 PM]

quote: Original post by zzzomed
If we take away how the human dody physically works, then we will better understand how the AI will come to the same result in forming its words, or anything else that it has to do.

Please satisfy my curiousity regarding this: Are you stating that if we remove the physical aspects of how the human body acts then we can better understand how the AI will come to the same results? I''m a bit confused by your statement here.

I just read all the way through this post, wow some of the posters here are incredibly intelligent! Makes me feel humbled.

In regards to the original topic, True AI, how can someone (anyone) regardless of how brilliant program a TRUE AI? At best wouldn''t they only recreate their own perception of an intelligence?

I think that the ability to combine wisdom (knowledge), compassion (can you program THAT in?), intuition and reason into any given decision is beyond duplication. I would be highly interested in seeing someone program an algorithm that duplicates true human emotion, not emulates based on a set knowledge tree. (Get mad because he hit me, be happy because he hugged me). What about just being in a bad mood for no discernable reason?

Somewhere up above I saw someone posted that in games REALISTIC AI is the goal. I disagree, and I''ll use an example from Gaming Tips from the Gurus (Game Programmer''s starter Kit 5.0):
In Deus Ex there''s a place where you walk outside. A NPC sees you, raises his rifle and shoots you dead, perfectly realistic action/reaction. Not fun for the gamer. He will have died and not known why or where from. So instead what they did is had the "bad guy" go (not literally) gosh there''s the person I''m supposed to be looking for, I better raise my gun up and kill him, then he proceeds to do just that. It gives the player time to recognize the threat and react. As Timkin stated, for Games AI that CHALLENGES the player is the goal. Besides, how many people want to play a game that they absolutely can''t beat?

Ok my 2 cents worth (and then some),

Anaton
Flying Tigers CFSG

In my opinion there is nothing at all random about how and in what order a conversation takes place, sure, it will be affected by many factors, but that doesn''t make it random. Wether you say ''Hi'' or ''Hello'' may depend on how you were brought up, how your friends speak (and how they were brought up) how your teachers speak, who your speaking to etc. I think you could develop a conversational AI that would seem perfectly ''human'', but, other than the actual speaking/english knowledge etc. (physical aspects?) the AI''s conversation skills would become influenced by the people it spoke to etc. You would find that initially it would probably sound very similar to the people who created and tested it, because those would be the only people it would have had speaking contact with. But once you let it talk to a variety of people, it may pick up new words, slang etc. Change the way it carries a conversation, and retaining knowledge about the people it talks to will also change it''s conversational style. There is also another thing i think would be important for it to truly have decent conversations, and that is some sort of needs system. If the AI doesn''t want/need anything, you would be hard pressed to get it to do anything other than answer questions. If it doesn''t want anything it won''t bother saying anything (and people get together for a ''chat'' because they want companionship, or sometimes other alterior motives). So there is a lot to take into consideration when trying to build an AI that is capable of holding a ''human'' conversation, but i don''t think randomness, or even unpredictability (on the part of the AI) is one of them.....unless you get the AI drunk... (i''ve noticed humans sometimes say things when they''re drunk that seem to bypass the brain completely ). Even if you don''t conciously think about what you''re going to say, i think it''s a good bet that your brain knows what you''re going to say before it sends the messages to operate your vocal chords.
Anyway, like i said, just my opinion.