But the computing power of the most powerful supercomputers in 2001 are far outstripped by the supercomputers in use now...Prof Stephen Hawking notes in 'Universe in a Nutshell' that the computing power of any man made computer today is far outsripped by the brain of even an earthworm, nuff said.
Look up a C++ Reference and learn How To Ask Questions The Smart WayOriginally Posted by Bjarne Stroustrup (2000-10-14)
This whole thread is irrelevant. Intelligence in the phrase AI has nothing to do with raw computing power.
Sure it does. With infinite raw computing power you could perform an instantanious search in some insanely large look-up table that would always provide you the best output given the available input; thus creating for an infinitely intelligent being, a perfect agent.
But yes, moot points aside, AI seeks to focus on better ways of utilizing the means we do have today... well, in lieu of our first implementaiton of a Turing machine at least.
no, it wouldn't.
An infinite data set would take an infinite time to search, and it woudl still need to be trained.
The most powerful super computer of 2001 is outstripped by a high end desktop of today...
Last edited by abachler; 01-09-2008 at 09:44 AM.
ok one second rewind. A AI IS NOT INTELEGENT (dont care about mispellings). They are just programs. Nothing more. Maybe later they will be but for now they are jsut made by us. They can only process stuff they cant reproduce and they are software. If you take awayt eh file it is looking throguh it is useless. A human ont ehr other hand can learn without being programed.
maybe later we will make so advanced AI that they will reproduce by running machines to make computers and then download their AI program to that computer. When that happens though I will laugh realy hard. Maybe they will operate electric wheel chairs to get around. They might also be able to program later, Improve their software, maybe even learn. But that is still a long ways away so for now they are not intelegent OK?
Oh yeah and a computer cannot hold more information then the human brain and wint for a VERY long time. Just so you know an adult only fills about 4% of their brain. imagine 4% now think of all that a single adult knows. Times that by 25 and you will get A LOT of information. An amiount that aa computer will take a very long time until we have computers able to store that much memory. (think like 1000 tetrabytes).
IMHO for an entity to be intelligent it must possess some form of consciousness. Without awareness things can only 'seem' intelligent.
Its not the storage capacity that computers lack, there are systems now with 1000 terabytes of ram, its processing power and architectural organization that they lack.
I have nver heard of a computer with 1000 terabytes of ram. It may be possible to do it wiht a network of supercomputers but a single computer is way beyond our technological level.
So i guess that gets them up to the brains of snails? have you heard of Moores law? quantum computer development is the only thing that will allow for this kind of brute force power approach
Of course I have heard of Moore's law, but I think you missed robwhit's reply to my observation: "Intelligence in the phrase AI has nothing to do with raw computing power."So i guess that gets them up to the brains of snails? have you heard of Moores law? quantum computer development is the only thing that will allow for this kind of brute force power approach
Even if we take "AI" out of the equation, it seems to me that Hawking's statement means nothing in this context since we cannot delegate our computing tasks to earthworms and snails, but we can delegate many of them to the computers we have at hand.
Look up a C++ Reference and learn How To Ask Questions The Smart Way
i am absolutely in agreement with RobWhit, that was the point i originally (somewhat obliquely) tried to make with the reference to earthworms, and i also commented again about the brute force aspect shortly after, it is in some ways i suppose analogous to a friend of mine that argued with me until he was blue in the face saying that a chess computer could always beat a human because it would just search out every possible game permutation leading from any given move, firstly he would not accept that it is just not possible to do that, tooo many gazillion permutations, secondly he did not realise what a wasteful approach that would be, (including all the wrong outcomes in your search) it is a question of approach, a chess player does not play like that, so why should the computer? is that the best way to try and approximate our thought processes? it is clearly not, and has not been engineered in that fashion by chess playing routines. a person playing against many opponents in exhibition matches is not searching a lookup table (though in a sense this is true because of research and training) and thinking n moves ahead so much as analysing each position on each board as they come to it and thinking of the best move from that given position.
Even with a perfect agent, capable of coming up with the right answer in short time you will not have intelligence. Intelligence doesn't mean you have an answer to a question. Thats knowledge. Intelligence is more than knowledge, it's the power to apply knowledge when needed. Sometimes, Intelligence is not answering the question, but shutting up and saying "yes dear". When a computer becomes aware of situations, for example not harassing me with compiler errors before I had my first coffee in the morning, complimenting her on the new pair of shoes even though it had cost more then the whole friggin' computer or not showing that flashy McBurger commercial when I'm thinking about a new diet, then I will take that as a hint of intelligence.With infinite raw computing power you could perform an instantanious search in some insanely large look-up table that would always provide you the best output given the available input; thus creating for an infinitely intelligent being, a perfect agent.
Just imagine what a bunch of cool functions we could have with intelligent computers: gethostbyintuition() etc...
hth
-nv
She was so Blonde, she spent 20 minutes looking at the orange juice can because it said "Concentrate."
When in doubt, read the FAQ.
Then ask a smart question.
and true, laserlight, a nice observation, but i simply wanted to draw attention to the relative processing power here, it is too easy to think silicone is king, because it isnt.Even if we take "AI" out of the equation, it seems to me that Hawking's statement means nothing in this context since we cannot delegate our computing tasks to earthworms and snails, but we can delegate many of them to the computers we have at hand.
Germanium is King, Silicon is just its easy to bed younger sister. GaAs is probably next in line for the throne.and true, laserlight, a nice observation, but i simply wanted to draw attention to the relative processing power here, it is too easy to think silicone is king, because it isnt.
40 years of research into silicon based semiconductors has taught us that we should have gone with germanium in the first place. Unfortunately, most of the research into Si doesnt apply to Ge.
Even neural networks arent AI, they are just linear adaptive filters capable of being used in image recognition and pattern classification. The application still has to be programmed how to use them or they are useless.
Last edited by abachler; 01-11-2008 at 09:41 AM.
