just what it says.. I just recently had a discussion with someone that ment that it was possible to make a "real" AI that would process and make own decisions based on nothing..
I mean, that it's not possible to have a AI to "think" on it's own. A AI cannot have a fantasy, for example having a word, and know the meaning of the word. and describe the word with other words, using it in different frases. And most of all making own words, which describes other things. A AI does only draw conclusions on things it already knows, right? I am not sure if you know what I mean, but I believe it's not possible to make a ai close to a humans intelligence..
in short, my opinion is: A machine can only know, it cannot understand.
Can anyone help me out here on this issue, explain some facts why it would or wouldnt be possible for a AI to work like this.?
at least not with todays technology
THanks in advance
Kenneth Wilhelmsen
Download my little game project HERE
--------------------------
He who joyfully marches to music in rank and file has already earned my contempt. He has
been given a large brain by mistake, since for him the spinal cord would fully suffice. This
disgrace to civilization should be done away with at once. Heroism at command, senseless
brutality, deplorable love-of-country stance, how violently I hate all this, how despicable and ignoble war is; I would rather be torn to shreds than be a part of so base an action! It is my conviction that killing under the cloak of war is nothing but an act of murder
Edited by - kenwi on November 3, 2001 2:38:24 PM
Anything like a "real" AI.
Started by kenwi, Nov 03 2001 07:36 AM
93 replies to this topic
Sponsor:
#2 Members - Reputation: 1114
Posted 03 November 2001 - 10:53 AM
quote:
a "real" AI that would process and make own decisions based on nothing..
You dont make decisions based on nothing. You need stimulus (for choices) to make decisions.
quote:
I mean, that it''s not possible to have a AI to "think" on it''s own.
If it looks like it''s thinking on its own, is there really a difference?
quote:
A AI cannot have a fantasy, for example having a word, and know the meaning of the word. and describe the word with other words, using it in different frases. And most of all making own words, which describes other things.
Why not? A word in a human sense is an oral representation of something we can recall from memory. Save an object''s most notable features to memory (to save space just like we humans do) and assosiate with a sound and now you have somewhat of an understanding of that object. Same can be done with actions or specific features. The hard part would be having the AI choose which words to use at what time (which we all do sometimes anyway) in order to convey a specific idea.
quote:
A AI does only draw conclusions on things it already knows, right?
So do you. If I talk about a Snorgoflax, then you''ll have no idea what I''m talking about.
quote:
A machine can only know, it cannot understand.
I disagree. What is understanding but fuller form of knowledge? If I say to the machine "The chicken crossed the road" and it now knows that the chicken has crossed the road and can reference that information on its own, does it not understand?
I think that if nature can create consciousness from a pile of grey matter, we can do the same.
Invader X
Invader''s Realm
#3 Anonymous Poster_Anonymous Poster_* Guests - Reputation:
Posted 03 November 2001 - 11:55 AM
The problem isn''t getting the AI to remember (store) facts and knowledge; it is getting it to understand relationships between them.
A computer system can store all sorts of fact, and even store some relationships between them. But it can''t intuitively figure out new relationships that humans can (at least as far as I have heard).
If they could do that, we would have SkyNet already, and Terminator 2 would have happened.
A computer system can store all sorts of fact, and even store some relationships between them. But it can''t intuitively figure out new relationships that humans can (at least as far as I have heard).
If they could do that, we would have SkyNet already, and Terminator 2 would have happened.
#5 Members - Reputation: 307
Posted 03 November 2001 - 02:49 PM
quote:
Original post by Invader X
Why wouldn't they be able to? What makes us so special?
Invader X
Invader's Realm
I think it's conciousness. We're more than just a bunch of neurons and brain cells, but a computer is just a bunch of ones and zeros. But this opens one huge can of worms
A CRPG in development...
Need help? Well, go FAQ yourself.
Edited by - Nazrix on November 3, 2001 9:50:17 PM
#6 Moderators - Reputation: 1087
Posted 03 November 2001 - 03:04 PM
quote:
Original post by Nazrix
We''re more than just a bunch of neurons and brain cells, but a computer is just a bunch of ones and zeros. But this opens one huge can of worms
Can you prove that? If you can, you could probably apply the same argument to computers
. [Resist Windows XP''s Invasive Production Activation Technology!]
#7 Members - Reputation: 1114
Posted 03 November 2001 - 05:58 PM
quote:
We''re more than just a bunch of neurons and brain cells
What else are we?
Invader X
Invader''s Realm
#8 Members - Reputation: 142
Posted 03 November 2001 - 06:02 PM
quote:
Original post by Invader X
--
We''re more than just a bunch of neurons and brain cells
--
What else are we?
Invader X
Invader''s Realm
Come on, we''ve all seen Final Fantasy (the movie).. *grin*
#9 Members - Reputation: 307
Posted 03 November 2001 - 06:46 PM
quote:
Original post by Null and Void
Can you prove that? If you can, you could probably apply the same argument to computers
[Resist Windows XP''s Invasive Production Activation Technology!]
Okay you have a point
How about this: We have the ability of self-conciousness. The neurons in our brain are able to be aware of themselves and ask why they exist among other questions.
I guess I can''t prove that computers and animals don''t do that though
#10 Members - Reputation: 307
Posted 03 November 2001 - 06:48 PM
quote:
Original post by Invader X
What else are we?
Invader X
Invader's Realm
That's where you delve into religion or lack thereof. That's where the "can of worms" is opened
Okay...everyone can thank Nazrix for causing the topic to go compltely off-course
A CRPG in development...
Need help? Well, go FAQ yourself.
Edited by - Nazrix on November 4, 2001 1:49:31 AM
#11 Members - Reputation: 100
Posted 04 November 2001 - 04:23 AM
quote:
Original post by Nazrix
How about this: We have the ability of self-conciousness. The neurons in our brain are able to be aware of themselves and ask why they exist among other questions.
But a real AI should ultimately have knowledge of its own internal mechanisms, its own physical structure, and the ability to examine its own memory space. Of course, humans didn''t always know what was in their brains or how they worked. We still don''t know entirely how they work.
As for consciousness, if we actually knew how and why consciousness exists, we wuold be better able to answer whether a computer could have consciousness.
Regarding an AI and knowledge of its own internal mechanisms, let''s look at a desktop computer. If we are ever to interact at an intelligent level with a desktop computer, it stands to reason that an AI residing on your desktop would have knowledge of all of the following:
- It should know that it''s software in ram and on a HD in a machine on your desk in a particular room in a living space in whatever town you live in whereever you live in the world.
- It''s composed of ones and zeros and these form a sequence of instructions.
- It should have an understanding of how the file system on your system works. Not just a program which gives you dialog boxes to browse files or is able to pull up directories, but a fundamental understanding of what a file is (a sequence of bytes) what recursive data structures are (the directory tree), what lists are (the lists of files in a directory), what a picture is, what file operations are (not just able to do them), what something like DELETE really means, etc.
- It should know what time is. I''m not talking about a clock or counter. I''m talking about time, its relativeness, its value, and the relationship between intervals of time then, now, and in the future. It should know what nighttime is, daytime, when you likely slept, when a long task of executing something might no be appropriate, etc.
- It should know what a keypress is, and should be able to tap into the rhythem of your keypresses, effectively assisting in you in your activities. But in a comprehending way, as the synergies of all of its knowledge work together.
- Much more...
I''ll give you an example of something that is NOT an AI, but rather something somewhat useless which masquerades as an AI and goes a long way towards convincing people that AIs will always be stupid: Microsoft''s personal assistants in packages like Word 2000. These little creatures embody none of the concepts I listed above. They have absolutely no understanding of what you are, what you are doing, what they are, or what the computer is.
___________________________________
#12 Anonymous Poster_Anonymous Poster_* Guests - Reputation:
Posted 04 November 2001 - 05:05 AM
Bishop_pass, you just LOVE lists, don''t you?
#13 Members - Reputation: 1114
Posted 04 November 2001 - 07:30 AM
Dont forget, that AI needs to want things that benefit it. If you try to delete the AI''s program or one of it''s main files, it would be able to and would resist you by any means possible.
Invader X
Invader''s Realm
Invader X
Invader''s Realm
#14 GDNet+ - Reputation: 438
Posted 04 November 2001 - 12:52 PM
quote:
Original post by Invader X
Dont forget, that AI needs to want things that benefit it. If you try to delete the AI''s program or one of it''s main files, it would be able to and would resist you by any means possible.
true... SkyNet didn''t nuke the Russians until they tried to unplug it...
--- krez (krezisback@aol.com)
#15 Members - Reputation: 1114
Posted 04 November 2001 - 07:03 PM
Also, by wants I dont mean automatic wants like David from AI had. He was very fake and wouldn''t pass for conscious at all. I mean wants which directly benefit the AI in some way and are thought about before they are wanted.
Emotions could be emulated in a similar beneficial/not beneficial break down making the difficult part of the AI the acutal analytical portion.
Invader X
Invader''s Realm
Emotions could be emulated in a similar beneficial/not beneficial break down making the difficult part of the AI the acutal analytical portion.
Invader X
Invader''s Realm
#16 Members - Reputation: 134
Posted 04 November 2001 - 07:30 PM
Has the idea of consciousness ever been explained? I''m not too knowledgable on this topic, so I''m just wondering if there is a certain configuration of neurons which will allow "sight", in a sense that it is.... like what I see, and thought like I think.... stupid English, not having the words I need to express myself
Trying is the first step towards failure.
Trying is the first step towards failure.
#17 Members - Reputation: 546
Posted 04 November 2001 - 10:11 PM
quote:
Original post by ragonastick
Has the idea of consciousness ever been explained? I''m not too knowledgable on this topic, so I''m just wondering if there is a certain configuration of neurons which will allow "sight", in a sense that it is.... like what I see, and thought like I think.... stupid English, not having the words I need to express myself
It''s tricky, but it''s possible to find out which parts of the brain map to different functions. For example, you can find out which part of the brain is responsible to sound by giving your subject some sort of sound stimulus (like ringing a bell) and seeing which parts of the brain become more active.
codeka.com - Just click it.
#18 Members - Reputation: 864
Posted 05 November 2001 - 12:52 PM
Just a couple of points I'd like to add:
Original post by Nazrix
Answer me this: which neurons in your brain are currently involved with interpreting the characters you are currently reading on this screen?
Our brains are NOT aware of their internal states, not by any stretch of the imagination. In fact, experiments seem to suggest that our brains are very delicately balanced complex systems that have evolved to self regulate themseleves. If you throw the balance out by just a small amount, the system collapses either into chaos (leading to various mental disorders of varying degrees) or to a shutdown of the control processes of the brain and ultimately death.
Out of this dynamic process emerges cognition and consciousness. There is a large and growing group of scientists that believe that animals have consciousness as we humans do and yet do not display the same cognitive abilities as us. This may suggest that consciousness and cognition are not as highly correlated as we once thought!
On the issue of whether something 'looks' intelligent as opposed to 'being' intelligent', might I suggest that interested readers obtain some of Stevan Harnad's papers on the turing test, modified turing test and the symbol grounding problem. They're available online.
Ultimately, we must consider that if something looks intelligent (through its actions/behaviour and, if possible, through discourse with that agent) then morally we must treat it as intelligent and accord it the same rights we expect as an intelligent agent. For this is exactly what we do with each other. We have no other means of ascertaining whether another human is intelligent other than from observation and expection: the observation of their behavior (and relating it to our own) and the expectation that since they are physiologically similar to ourselves and that we believe we are intelligent, then we accept that they are intelligent. It is the whole presumption of the need for physiological similarity before we can accept something as intelligent that makes it difficult for us to believe that other agents - be they organic or inorganic - could be 'intelligent'.
If we co-existed on this planet with another species that was intelligent and that we could readily communicate with, then I doubt very much that so many people would be so quick to assume that machines could not be intelligent.
Regards,
Timkin
Edited by - Timkin on November 5, 2001 7:59:37 PM
Original post by Nazrix
quote:
How about this: We have the ability of self-conciousness. The neurons in our brain are able to be aware of themselves and ask why they exist among other questions.
Answer me this: which neurons in your brain are currently involved with interpreting the characters you are currently reading on this screen?
Our brains are NOT aware of their internal states, not by any stretch of the imagination. In fact, experiments seem to suggest that our brains are very delicately balanced complex systems that have evolved to self regulate themseleves. If you throw the balance out by just a small amount, the system collapses either into chaos (leading to various mental disorders of varying degrees) or to a shutdown of the control processes of the brain and ultimately death.
Out of this dynamic process emerges cognition and consciousness. There is a large and growing group of scientists that believe that animals have consciousness as we humans do and yet do not display the same cognitive abilities as us. This may suggest that consciousness and cognition are not as highly correlated as we once thought!
On the issue of whether something 'looks' intelligent as opposed to 'being' intelligent', might I suggest that interested readers obtain some of Stevan Harnad's papers on the turing test, modified turing test and the symbol grounding problem. They're available online.
Ultimately, we must consider that if something looks intelligent (through its actions/behaviour and, if possible, through discourse with that agent) then morally we must treat it as intelligent and accord it the same rights we expect as an intelligent agent. For this is exactly what we do with each other. We have no other means of ascertaining whether another human is intelligent other than from observation and expection: the observation of their behavior (and relating it to our own) and the expectation that since they are physiologically similar to ourselves and that we believe we are intelligent, then we accept that they are intelligent. It is the whole presumption of the need for physiological similarity before we can accept something as intelligent that makes it difficult for us to believe that other agents - be they organic or inorganic - could be 'intelligent'.
If we co-existed on this planet with another species that was intelligent and that we could readily communicate with, then I doubt very much that so many people would be so quick to assume that machines could not be intelligent.
Regards,
Timkin
Edited by - Timkin on November 5, 2001 7:59:37 PM
#19 Members - Reputation: 307
Posted 05 November 2001 - 04:02 PM
My point was more a matter of:
as far as we know, animals do not go around pondering life''s questions or other such activities. The most they *seem* to think is "I''m hungry" or "I''m happy". Yet, wouldn''t you say most animals are more autonomous than computers are so far?
Is it "possible" to make computers have the level of intelligence that we as humans do? Probably. But I don''t think we understand enough about ourselves to do that at this point in our technological status, do we?
A CRPG in development...
Need help? Well, go FAQ yourself.
as far as we know, animals do not go around pondering life''s questions or other such activities. The most they *seem* to think is "I''m hungry" or "I''m happy". Yet, wouldn''t you say most animals are more autonomous than computers are so far?
Is it "possible" to make computers have the level of intelligence that we as humans do? Probably. But I don''t think we understand enough about ourselves to do that at this point in our technological status, do we?
A CRPG in development...
Need help? Well, go FAQ yourself.
#20 Members - Reputation: 307
Posted 05 November 2001 - 04:06 PM
quote:
Original post by Timkin
Out of this dynamic process emerges cognition and consciousness. There is a large and growing group of scientists that believe that animals have consciousness as we humans do and yet do not display the same cognitive abilities as us. This may suggest that consciousness and cognition are not as highly correlated as we once thought!
That''s pretty interesting. Actually if you look at humans from a very objective view, and a very macro view we probably look like bees or ants. We travel to the same basic places each day, we travel to get food. We mate and reproduce. We fight each other for resources.
A CRPG in development...
Need help? Well, go FAQ yourself.
