Where are we from ?

Sylvain:

I think there will be still some time before a computer is able to think, and probably even more before acomputer is able to CREATE.

On of the main differencies is that a computer can be considered as perfect and limited. It is limited to 0 and 1 and never goes out of that, or if it goes, it crashes...

Human are not limited to that, and the imperfection of the brain makes it very sensible to a lot of things. You can remember wrong things, you can select what your memory retains, and you memory can retain things that you are not aware of, and that can suddenly come back ion a middle of something else. The size of the brain evolve (it grows and then shrinks), and the connections between neurons changes a lot, for some time no real reason. All these are probably some of the main reasons the brain is creative (and not only human brain), and a reason why the actual type of computer won't be able to be CREATIVE, as long as I know.

We can go on long on that way, an the fact to have computer more powerful and complex are probably not enough to have really thinking machins.

Qui sait....

Didn't the french/british colonies become 'Canada east/west' with the Act of Union in 1840? Funny I don't recall much time spent on canada in history classes ;) And it wasn't a revolution it was just infighting between the french and british colonies as I recall it. I think the british eventually merged all the colonies but that's just off the top of my head. Guess I'll read up on it tonight ;) Unless some canadian wants to enlighten us.

Quote:

---

Originally posted by Mick_2002
Didn't the french/british colonies become 'Canada east/west' with the Act of Union in 1840? Funny I don't recall much time spent on canada in history classes ;) And it wasn't a revolution it was just infighting between the french and british colonies as I recall it. I think the british eventually merged all the colonies but that's just off the top of my head. Guess I'll read up on it tonight ;) Unless some canadian wants to enlighten us.

---

I was talking about the American revolution, who was against England. Canada was considered a british colony until 1867 when the conferedation occured. So before the american revolution and after 1760 when the seven year war ended, which defeated and eradicated New France, both "americans" and "canadians" were loyal subject to the King of England.

After the end of the seven years war, referred in american history as the french and indian war, british colonies moved north to Lower and Upper canada, mainly Upper Canada though, Lower Canda was already filled with french catholics(!).
Lower Canada is what is now known as Quebec, Upper Canada is what is now Ontario.

Lower and Upper Canada joined to form Canada. I'm not going through all the changes in territories of Canada, they are too numerous.

So anyway, Canada stayed loyal to the crown, the USA got their independance. I'm not saying Canada didn't exist as a separate entity before the war of independance, I'm saying both people were from the same background, hence have some of the same celebrations.

Of course, this is a generalization, please, no flame war about how Canada is different from the USA, please! :)

Of course, when I say "canadians" I'm talking about canadians from british roots. Not "canadiens", of french descent. We have nothing to do with any of that. We were just in the way, that's all ;)

ThanksGiving, in french: Fête de l’Action de Grâces

For that AI thing, I've already see something interesting about that. It said that we can't create an AI with a bunch of programmers and big databases. You know, for example, converting "what your ear" to "the meaning of the words". It would require a lot of works and would be as crap as <<censored product name :D >>.

For example, you could create a strong AI that answer any question in any langage, but in 100 years, langage will have changed... probably..., the programmers would not know how to answer "Where is God?" or that kind of question, so it would be as generic as Microsoft Word Trombine sensitive help.

The ANSWER to these problems is to create a THINKING AI, rather than an AI. A simple program that know the basis and learn, the same way a child learn. Using "computer", we could transfer that learning from AI generation to others AI generation (Humans child need to learn everytime... :D).

It has already been done, a sample, sample and very sample AI that learn and share it's knowledge. It was little robots with wheels. They started knowing nothing except to move. They live to "deplace themself". At first, they hit walls, other robots, objects, and cannot move properly, but with a "thinking AI", they learn quickly to avoid wall and objects, and that create "moving robot"

Another example is "the following robot" (cat and mouse). Two robots are created. One is "the cat", the other "the mouse". The mouse have only one purpose, flee from the cat that have only the purpose to grab the mouse. This was used to demonstrate the learning ability and the sharing ability from generation 1 to generation 2. The first generation of cat and mouse were extremely stupid, moving like drunk men, not able to do anything good. Each generation was better, at a time, the cat was very close to the mouse and the mouse was curving to flee away.

Do not ask me the exact "how they are done", I see that on TV, pretty interesting IMO :)

Anyway, having thinking AI will be in a long long time, and as you have stated, they might turn against us, and this is not a joke. You know, to have such intelligent being means we CANNOT program it like we create .EXE in VB, you know, it requires high technology like the learning I just state or the biological-electronic learning chipset (in development, this is an electronic chip where biological cells grow up, impressing if it works). So, if we have intelligent being that share knowledge just by plugging an USB cable from one to another (:D). They will quickly learn how to repair themself, how to create others, how to have a "metal frame" (if humans did not granted them an enough solid one), so they will know that they are far superior than human... (Let the bodies sleep and grow)

A scientific (again at this TV show I think) said that the next evolution step is android (you know, human with implants). He was saying that "since we have the knowledge to do it, since God grant us the knowledge, then it is the next step". you know, having bionic eyes and that kind of Sci-Fi stuff is, today, not so Sci-Fi ;)

JeffB

Interesting :)

Yeah, there is this crazy professor in the UK called Kevin Warwick (I think it's Warwick) who has already had a few implants himself. Some people want to implant radio emitters into their children so that they can trace them in case they get lost or more importantly abducted.

Actually AI is a very generic term. The AI field that uses databases is mostly called expert systems. They are very useful in themselves. The ability to make generalizations and learn from experience, handle special cases etc, is what makes the AI part of it. A big database in itself that doesn't learn by itself is not AI.

Quote:

---

Originally posted by Yves M
A big database in itself that doesn't learn by itself is not AI.

---

Well, as you said AI is generic, but a program that answer any question using a big databases of word is still an AI (not a thru one though ;) ) Game BOT (you know, the ennemy) use AI, some are hardly coded (with IF statement: IF someone near you, ChangeCap(), Attack() ), and some are adjusting using complex IF statement, so they will "think" at what can happens if Me.Attack() = DEAD then Me.Flee() Chess AI is still AI, there is Chess AI that can learn, but even those who don't learn are Artificial Intelligence. ;)

JeffB

Yeah, well the term is much too broad. On the other hand there are the neural nets that learn from experience and don't have much prior knowledge themselves.

A chess program is really most of the time only a calculator. Although recent programs do include more advanced concepts about the game, like recognizing positions, formations, strategic objectives instead of calculating everything.

And AI from Games is most of the time laughable. But they are getting better as well, although I haven't seen any game yet where the machine improves as it plays games against you. Decisions are usually randomly picked from a set of best decisions and no knowledge is accumulated. It would be very interesting to play games that would learn to play better.

Ok, I have myself programmed a Tic-Tac-Toe that improves its play as you play against it, but well, that's nothing hard ;)

I've alredy read a book about learning AI in games (Tic-Tac-Toe learning AI ;), then Chess learning) It is possible, but games retailers do not share a lot (IMO) about their AI. They just say the AI is better than before, but I've once heard that Age of Empires (2? probably) AI was learning from your play. Don't know "how", if it just "Be more aggressive, be more defense, do not care at all", but I know that a good learning AI would learn from it's bad experience, for example, when he lose, he analyse the game and try to find the errors he has done, and then, he'll learn that he was having 5 siege weapons that did nothing in the combat (or he can do that at run-time while you play it).

Hard to implement? Yes probably, but you can still do a little one. Playing StarCraft for example, how many times the ennemy has send me a "Scout", that did absolutely nothing but die before even try to turn around (Scout cost many ressources by the way). A simpler AI would analyse all his "combat" and his "units lost", in that case, he would not send another Scout there (at least send something cheaper to scout the area) :)

You're right AI in games in general are just plain script. They cast a magic that bounce (thanks to your magic shield) right into them. And then, they cast it again, and again, and again, until they are dust. What's the!? It is an "AI", but a "good AI" would quickly find that "The magic XXX is worthless" by simple calculation. (last example from Baldur's Gates and Icewind dale series)

I know it is possible to find such a thing because the learning Chess program CAN be done, and it is quite the same. A bad move is when you move and you lose the piece (for example). When you throw and arrow and see it bounce on a magical Shield to hit you back, it IS a bad move :D :) (you lose more than your opponent, please sheathe that bow and take your dagger ;)) (last example from Baldur's Gates 2) :D


JeffB

I've been playing Warcraft III recently and even though the AI is quite strong, it is also quite limited. It's a bit the same thing as your example with the magic shield. They go out to kill a flying dragon and have only units that can attack ground targets :rolleyes: They will stay there until all their units are destroyed :p

I'm really waiting for Master of Orion III though, apparently the AI should be quite interesting.

The first amazing game AI will be the one that will be able to make up tactics without help from the programmer. I think the technology is there, but games have been too focused on graphics instead of AI. It's getting better though :)

Geez guys! There were like 50 or more posts since yesterday! Slow down for me!

Quote:

---

Originally posted by Yves M
The first amazing game AI will be the one that will be able to make up tactics without help from the programmer.

---

The ultimate challenge would be this. You have some way to tell the computer every rule of some game, how to play and make a move, and nothing more. It would then take this into consideration in some algorithm and come out with the best move. It has been done before, only context sensitive. For example, everyone knows about Deep Blue, IBM's mastermind. It could analyze any chess position and make a better move than any human chess player on the planet. If we could extrapolate this process to include any game at all, it would be incredible. Of course, there are many, many, many problems first which is why it hasn't been done yet. First of all, some games have a random factor, like a card game, and some have unknown information, also like in a card game where you can't see your opponent's hand, very much unlike chess. Also, some games require an interpretation of your opponent's posture and expression, such as poker. Along with all that, the immense data and calculation time it would take to be able to learn any game and play it well would be through the roof. Aside from these minor deterrents, such a program would be the end-all of AIs.

Well, I don't really expect this kind of AI to be in our hands anytime soon. I don't think that it's bad that the programmer gives the machine some strategies to work with, some hard rules of what is bad, what is definitely good etc. If you had to train the game before it was interesting to play then it wouldn't be much fun either.

Moreover, think about computer games like Warcraft, RPGs, etc. They would reap the real benefit from a learning AI. I mean having a chess program learn as it goes along is a nice academical exercise, but it's not much use (or fun ;)).

Actually in RPGs having a strong AI would be tantamount to creating a real virtual world. I've played some MMORPGs (Massively multiplayer online role-playing games) and they are great since there is that "human" factor in most avatars you encounter. Now think about an RPG where the non-player characters would have a strong AI, their own motivations, their learning, their decisions which are not preprogrammed by the game designer. I'm sure we'll see some of these things in the next ten years. Maybe not quite as evolved, but still, we'll get there ;)

Quote:

---

Originally posted by Yves M


Sunny, but a bit cold (only 25 degrees :p)

---

:D :D :D ...and today? :D :D :D

Good morning all,

If AI in games were better (in games like warcraft, starcraft...), i'm not sure the games would be better, or at least interresting. Actually when you play against the computer, sometimes it is hard to win. Of course, computer is a bit stupid and from a game to another, it always has the same rules and you can easily see what it's going to do.
But now, imagine you play against a computer which has the same intelligence than you, PLUS all the advantages to be a computer..then you do not have any chances to win...you stop to play the game..ahah.

Maybe one day, scientist will understand how works the human brain. Exactly how it works. And this day, they'll try to make a software which do the same thing. It will be the first artificial brain. So, if the fact to think that if you've got a "complex brain" means you can have one mind and to be intelligent, then that will be the start of i don't know what..maybe good, maybe bad.
Our brain is biological, why not the same thing with electronics ?

Before that, i think we will have some "androids" like JeffB was saying. No more glasses, we will have some kind of i don't know what in our eyes...no more problems of ears...blablabla.

I've seen many many times on TV some reports about how to live longer, how to "not become an old human"...we always invent some new medics (medics? we say that in english ?) to improve and make your body live longer and faster. Why? ok to live better since we die, but not longer ! what about the human children. I hate scientists who invent only to be proud of their work, to have their name in a book one day. They don't think about future generation. I say all this, because it's the same problem as AI:
do we need to invent and search all this things? why not stop evolution before big problems occur.

This is easy to say that. Not to do it, i know.

Good day for all of you.

Sylvain.

Quote:

---

Originally posted by Sylvain
But now, imagine you play against a computer which has the same intelligence than you, PLUS all the advantages to be a computer..then you do not have any chances to win...you stop to play the game..ahah.

---

That's TRUE. The advantage the computer has is that it can give 10 orders (or much more) when you only have time to give one. That's why having game with really strong and learning AI would require a way to slow its actions to make them look more than what humans can do. That's at least the problem with real time games.

For games like chess, ... then a real AI is defintely an advantage.

Quote:

---

Our brain is biological, why not the same thing with electronics ?

---

It may not be that simple. Or may be it is. Human brain can have undefined behaviour that an electronics could hardly have.

Quote:

---

I've seen many many times on TV some reports about how to live longer, how to "not become an old human"...we always invent some new medics (medics? we say that in english ?) to improve and make your body live longer and faster. Why? ok to live better since we die, but not longer ! what about the human children. I hate scientists who invent only to be proud of their work, to have their name in a book one day. They don't think about future generation. I say all this, because it's the same problem as AI:
do we need to invent and search all this things? why not stop evolution before big problems occur.

---

Most people want to live longer, just because they don't want to die. That's a good enough reason why medecine that makes people live longer exists.

I am sure we could find a lot of good reasons to want real AI, and not only for games. But it's still the morning and my brain is a little slow :rolleyes: