Is artificial intelligence possible ?

[IsthatyouJohnWayne 0]

One of the problems with AI 'learning' its way to increased intelligence is how damn long it would take to get anything but a retard

Compared to a baby or young child, artificial systems seem to me to 'learn' in a very inefficient way, at least with regards to forming reasonable assumptions of a complex or general nature about 'the world' .

I seem to remember that partly for this reason there is an AI  project where a guy is essentially trying to codify the laws of common sense (in as much as is possible) by stating to a computer every common sense rule he (or his team) can come up with. I believe this is an example of the 'top down' approach to AI.

If it is helpful for AI learners to have goals then there is the problem of how one would precisely define the goal of acheiving intelligence? (im sure there are complex solutions)

It may be that we end up in the not too distant future

with AI that 'knows' many things and can learn somewhat similarly to a human and recognise patterns, can respond appropriatly to stimuli, can recognise objects, can negotiate terrain and situations fairly intelligently and to a limited but significant extent can speak with people 'intelligently' BUT that lacks the vital spark of genuine sentience. Still im willing to bet that many people would be willing to proclaim such a device 'intelligent'

[Major Fubar 0]

I think through extensive IF/THEN/ELSE type logic and an element of random number generation, a program could be written that mimics true intelligence so closely it would be hard to tell from the real thing.

Of course, it's hard to say when: might be next year, might be 200 years...

My 2 cents...

[bn880 5]

The above 3 posts is exactly what I'm talking about.

Problem: defining intelligence

If you are not willing to define intelligence, there is no point talking now is there. And so the question of the thread is "Is artificial intelligence possible ?" when intelligence is undefined, you are just going in circles.

For me intelligence is a response chosen based upon previously learned events. So I find many adaptive systems intelligent in a small sense. Such as the adaptive tree logging machines, they are intelligent to my understanding. No they are not capable of the level of human intelligence, but a tiny fraction.

[ralphwiggum 6]

IF/THEN/ELSE is a good start, but my personal opinion is that it will run to problem of massive operation.

IF/THEN/ELSE works for 2 choices. either the statement is true, or it is false. theoretically, one can deduce to the point where they get to the one part that is intended to be, but that takes a lot of iteration, since you have to go true/false statement a lot.

of course our PCs that we use to connect to this is already strong enough to do a lot of iterations, but i doubt if that will happen.

when BIS created me they added some creativity codes so that i look more like an individual person typing rather than regurgitating slowly what humans type.

[bn880 5]

IF THEN..? You have to start by looking at the type of communication between brain cells, which I suspect is very much analogous in the sense that every signal sent has a quantitative value based on it's strength and is acted upon accordingly by the recieveng cell(s).

In a digital system you are looking at (basically) on or off, 0 or 1. It's not a programming issue then, it's a hardware design issue to duplicate a brain as it will may some sort of use of analogous circuitry. To compensate for analogous signals digitally you are looking at enormous processing problems.

[Tovarish 0]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote (bn880 @ 21 May 2003,02:47)</td></tr><tr><td id="QUOTE">IF THEN..?  You have to start by looking at the type of communication between brain cells, which I suspect is very much analogous in the sense that every signal sent has a quantitative value based on it's strength and is acted upon accordingly by the recieveng cell(s).

.<span id='postcolor'>

Sounds like a State Machine .

But then if every state is analogous to a braincell.... ....You won't fit that on the chips we're using that's for certain . If you think about it that way that's some incredible processing power needed. But it sort of makes sense....something like the worlds largest collection of interconnected state machines for decision making - since it would be a damn slow brain if only one cell at a time was active -....and a sort of uber hash-table for memory management...and you might have something resembling an animal brain.

The only downside to that approach other than the sheer computing power you'd need to run it would be the half a dozen centuries it would probably take to design - nevermind debug.

*edit* Since I get a feeling that for many people what I've just said is mostly gibberish....basically I'm coming to the conclusion that trying to duplicate the structure of a brain with software/digital logic is damn near impossible...Maybe the only practical approach is to keep working on these "self teaching" machines and hope for evolution .

[IsthatyouJohnWayne 0]

Major Fubar-"I think through extensive IF/THEN/ELSE type logic and an element of random number generation, a program could be written that mimics true intelligence so closely it would be hard to tell from the real thing."

Simple text programs have been written since the seventies that acheive that aim (that is mimic intelligence in a variety of ways well enough to convince or trick the average person into thinking it was human at least for a period of time). Thats not for me the measure of true intelligence though. If it is merely -hard to tell- from the real thing then it is self evidently NOT the real thing. Perhaps you really mean that it might essentially become the real thing without us knowing it

bn880- its true that the word 'intelligence' is quite nebulous and can have many connotations.

"For me intelligence is a response chosen based upon previously learned events. So I find many adaptive systems intelligent in a small sense. Such as the adaptive tree logging machines, they are intelligent to my understanding. No they are not capable of the level of human intelligence, but a tiny fraction."

Yes according to that definition any such system would constitute artificial intelligence (,a level thereof).

But seeing as artificial intelligence by that definition is already a reality, the question of this topic 'is artificial intelligence possible' would seem to be a rather stupid one. However it is clear that the term 'artificial intelligence' tends to carry certain associations of consciousness, sentience and existential awareness. I agree that artificial stupidity already exists. We might now fairly well be able to make a body with sensors and create something to match the brain of a worm or small insect but i would not call worms or insects individually especially intelligent. This topic does indeed raise a number of interesting questions. Does a scale of intelligence arise simply of knowledge retention and the ability to learn, with a certain point of sophistication and complexity at which consciousness naturally occurs? Will we ever even know for sure when that point is reached ? Perhaps, perhaps not.

Is human type higher intelligence the only recognised or legitimate type?

I think it must be as we cannot conceive of a non-humanesque form of higher intelligence and we would thus be unable to recognise it (probably).

Perhaps we will not be able to tell when we have a sentient AI and when we have only a very convincing fraud. The Turing test is one (elderly) possibility for an AI test.

[ralphwiggum 6]

IIRC, brain works so that specific part is combined with others to make a working combination.

Personally, I doubt if a human made machine and human made code can fully compete with human's brain. after all, humans brain had millions of years to beta test and still will.

[Tovarish 0]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote (RalphWiggum @ 21 May 2003,05:27)</td></tr><tr><td id="QUOTE">IIRC, brain works so that specific part is combined with others to make a working combination.

Personally, I doubt if a human made machine and human made code can fully compete with human's brain. after all, humans brain had millions of years to beta test and still will.<span id='postcolor'>

My thoughts exactly. Get the AI equivalent of a microbe, and IMHO we're already there with neural networks...and from that point on I suppose we could try to speed up evolution by providing appropriate stimulus....but it could take forever before there's a machine equivalent of an animal brain.

I'm certain that we'll have machines that mimic the behaviour of an animal well enough that you could put a realistic skin on it and it may fool people for a prolonged period of time. Maybe soon after the same could be said for machines that mimic humans - but would they have our reasoning ability? No.

Who knows, by the time an AI exists that functions on the level of the human brain, the human race itself may have gone the way of the Neanderthal.

[bn880 5]

Well so maybe we aren't going to re-create the brain quite so digitally, rather use a biological / digital combination, biocomputing.

Yes I think you are seeing the exact problem I was reffering to Tovarish. Compensating for all the ranges in a single signal and processing each cell as a miniature state machine. (simultaneously)

@John Yes the thread title is a bit silly since we know it is, and that is how I define intelligence. Animal or human intelligence is something much more or simply larger by some huge factor.

[Oligo 1]

Somebody once said that if we someday managed to create an intelligent machine, we wouldn't realize it, because we are not intelligent ourselves.

[Nagual 0]

Here's something some of you may find interesting, taken out of context, but still:

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote </td></tr><tr><td id="QUOTE">In this respect, understand that both intelligence, as well as awareness, imply consciousness. The problem in trying to understand this lies in the fact that, generally speaking, man is very careless in his understanding of the words he uses. Consequently most of humanity cannot distinguish between the terms awareness, intelligence and consciousness, simply because these three terms are usually assumed to be synonymous. Yet in actual fact, although the three words are closely allied, since they are interdependent, there is nevertheless a vast difference between the true implications of these terms. The word 'intelligence' comes from the Latin word 'intelligere', meaning 'to choose between'; whereas the word 'awareness' comes from the Latin word 'vereri', meaning 'to be fearful'. We therefore have, on the one hand, a state of consciousness which brings about discernment, and on the other hand, a state of consciousness which recognises the necessity to be alert. Both, however, are expressions of consciousness, a word which comes from the Latin word 'conscius', meaning 'sharing knowledge'.

Nagal Mares

<span id='postcolor'>

Anyhoo, i think they will make AI (as it is Artificial, not actual intelligence), but prolly only when some good chemical based computers are made.

[-radkeff- 0]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote </td></tr><tr><td id="QUOTE">Anyhoo, i think they will make AI (as it is Artificial, not actual intelligence), but prolly only when some good chemical based computers are made.<span id='postcolor'>

I can agree with that...

[Supah 0]

I thought this was about blond women changing hair colour (gf did that once)

[tracy_t 0]

Of course machine artificial intelligence is possible. after all, in a reductionist's eyes, what are we but biological machines?

All we have to do is fully understand the processes of the brain

Bn880's question about what intelligence actually *is* has been debated since time immemorial, and I dare say the argument will continue long after we are dead. There's a sobering thought for you.

Personally, I don't think I will see Robots in our lifetime. But then again I'm 90 years old. J/k

[-radkeff- 0]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote </td></tr><tr><td id="QUOTE">Yes according to that definition any such system would constitute artificial intelligence (,a level thereof).

But seeing as artificial intelligence by that definition is already a reality, the question of this topic 'is artificial intelligence possible' would seem to be a rather stupid one.<span id='postcolor'>

Yes according to that ai exists today would make the topic title stupid...but i have actually defined what i meant with artificial intelligence :

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote </td></tr><tr><td id="QUOTE">Will real artificial intelligence ever be possible ? By that i mean will computers ever think, learn and solve problems by themselfs ? I serously doubt it... <span id='postcolor'>

And That dont exist today...

Now to enhance the defininition of the intelligence im writing down some words that i think should be included when defining intelligence:

-Reasoning-

-Analysing-

-Language-understanding-

-Learning-

-Problem-solving-

Ofcourse even these are a vague definition intelligence and im sure more can be added to the list...But to point out what i meant is will humans ever create something that is as smart or smarter than the human brain ?

Edit: Or create something that can do all or nearly all the things on the list above

[Albert Schweitzer 10]

Absolut CPU

[Guest]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote (-radkeff- @ 21 May 2003,10:12)</td></tr><tr><td id="QUOTE"></span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote </td></tr><tr><td id="QUOTE">Anyhoo, i think they will make AI (as it is Artificial, not actual intelligence), but prolly only when some good chemical based computers are made.<span id='postcolor'>

I can agree with that...<span id='postcolor'>

Actually I'd go with quantum programming on this one. The sort of statistical/logical weirdness that goes on with those sorts of quantum states would seem to me to be the best chance of getting an AI (as vaguely defined above) happening.

In some ways I think intellegence relies on Self-awareness/consciousness. Otherwise it is just a tool. Even plants can communicate with each other (release gases etc), so a computer has a long way to go.

[bn880 5]

radkeff, so intelligence in your view is basically only a human quality. Even some animals have no intelligence then because they can not meet that criteria.

Artificial intelligence has existed for a long time already, just on a small level. It's pretty sad something has to take the shape or form of a n animals intelligence to be understood. But anyway, as always you can debate for ever since debating never solves anything.

[-radkeff- 0]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote </td></tr><tr><td id="QUOTE">radkeff, so intelligence in your view is basically only a human quality.<span id='postcolor'>

First of all i never claimed that...second i never claimed that my list was definite...however cant you have more or less of a skill ?..like 0.03 analysing, 10 learning and 7 problem solving ?

maybe a cheesy example but you get the point..

now lets debate

[IsthatyouJohnWayne 0]

Exactly Leone.

Actually there are many possibilities for the way computer technology may advance (benefitting AI) in the next century or centuries. Because we are approaching the limit of silicon technology, conventional microprecessors might soon become a major barrier in the advancement of computer technology. Even if you start stacking microchips into cubes to up the performance there is the problem of the scorching heat generated by such processors (requiring advanced cooling techniques or alternatively a room temperature superconductor). Short term solutions might include using gallium arsenide or silicon germanium instead of standard silicon to make circuits perhaps 10 times faster, using X-rays instead of laser beams to etch components (or beams of electrons or extreme ultra violet) but at some point when the distance between wires is too small electrons will simply leak between one and another thus ruining the circuit.

Some of the possible replacements/alternatives in the medium to far future include:

optical computers (perhaps with holographic memory)

bio molecular DNA computers using chemical code A T C G  (representing DNAs four nucleic acids)

Bionics attaching hardware to wetware (melding man and machine)

Quantum transistors (controlling resonance of a quantum device by raising/lowering the voltage; regulating electron flow through the device corresponding to 0 or 1)  

Lastly full quantum wave based Quantum computers making possible all kinds of astronomic wackiness thanks to 'qubits' that exist simultaneously as a one and a zero. But i personally think a full on quantum computer will be unfeasible for a very very

long time ( over a century or centuries+ ). Perhaps experimental examples may exist before then but it wouldnt be a normal consumer device for the home to play Operation Flashpoint 36 on like a home PC (at least not for an age)

, perhaps there would at first be a community Quantum computer that all of a community (or city)

could access and use simultaneously but thats getting into the confusingly distant future.

So who knows in which direction the next technological step

will be taken?

[Jinef 2]

This is quite interesting,

my definition of intelligence would be to synthesise information and then react accordingly as you think appropriate.

Thanks for that little tutorial Denoir... I am now more confused than i could possibly imagine, i sort of skimmed it. Maybe when i have energy i will find out what is 'going down' with all this programming molarchy

[Guest]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote (IsthatyouJohnWayne @ 21 May 2003,16:53)</td></tr><tr><td id="QUOTE">Lastly Quantum computers making possible all kinds of astronomic wackiness thanks to 'qubits' that exist simultaneously as a one and a zero. But i personally think a full on quantum computer will be unfeasible for a very very

long time ( over a century or centuries+ ).<span id='postcolor'>

I'd say 20 years at most. You already have functioning examples that can perform simple logical and math operations.

What is lacking today is the production technology that can mass produce the devices with sufficient accuracy. Also still a problem is the stability and sterility of the operating environment.

I'm confident however that it will be solved

[Guest]

</span><table border="0" align="center" width="95%" cellpadding="3" cellspacing="1"><tr><td>Quote (Jinef @ 21 May 2003,16:55)</td></tr><tr><td id="QUOTE">my definition of intelligence would be to synthesise information and then react accordingly as you think appropriate.<span id='postcolor'>

A bit of a cirular logic there

I think that what people most often think by "intelligence" is self awareness.

[IsthatyouJohnWayne 0]

I just noticed that Denoir, what i was referring to really was the point at which non experimental models become widely used and i think that will take a century plus (im anticipating some significant hurdles)

Anyway i think an interesting question is how all of this technology and 'intelligent' ai will change society (human society).

And will there still be people starving in Africa even as we create this artificial life in a laboratory?