I am doing a research paper for my 102 english class.
My topic is how can an A.I. be made that is intelligent by human standereds of an intelligent person?
Is there any good open source example of an A.I. that works well?
I know M.I.T is working on A.I. and robots but not all of there devlopment information is open to every one. (As far as I can tell)
Is there any basic A.I. programs that are open source that does not take a dagree in Computer science and math to understand?

Is there a good web site with jurnoles of expermints like they have for phsyc?
Is there any one person or group that I can contact to interview them on their ides of A.I. that work in that field?

You can create A.I. that is more intelligant than human,
simply by taking normal pocket calculator and augmenting it by living human brain.


Google & Youtube
"Michio Kaku AI"
He explains consepts AI in way that is easy to understand.
(note: most of AI`s that he talks are future AIs (100-500 years to future))


/Tyrian

Ps. tell me how do you think, and I will tell how to create A.I. :D

How about searching the web for information about computers that can play chess. I think that's the area where AI has actually reached a level where it can compete with humans. Other than that, AI is still far from being able to match human intelligence.
Neural nets are probably the closest thing to human intelligence as far as the basic mechanism of how they work compared to the human brain, you might want to research those as well.

With A.I. and chesss. It is a type of A.I. but it can also be seen that there are only so many spaces and so many piaces so it really just uses acomplicated algorith to find every move possible by genrlizing many things. Yes that is a type Artficale intelgince but will never be considered real intelgince by many.

I fell that they are trying to creat A.I. from the wrong direction.

Quote: Original post by kingpinzs
With A.I. and chesss. It is a type of A.I. but it can also be seen that there are only so many spaces and so many piaces so it really just uses acomplicated algorith to find every move possible by genrlizing many things. Yes that is a type Artficale intelgince but will never be considered real intelgince by many.

If you asked people in the fifties if a machine that could play chess better than any human would be considered intelligent, they would probably say yes. Once you actually do it, you know how you did it and now it just looks like "a complicated algorithm", and not "real intelligence". In that sense, you'll probably never be satisfied with AI, and the only reason you think people are intelligent is because you don't know how their minds work.

Quote: I fell that they are trying to creat A.I. from the wrong direction.

You can probably learn something by reading about the difference between "strong AI" and "weak AI". The people that are going in this "wrong direction" are actually getting certain problems solved.

If you can solve problems in a different way, I am all for it. Until then, I think you should be a bit more modest and try to learn. On the other hand, there is nothing wrong with thinking big.

Statistics and AI are _very_ related fields. Many branches of the sciences (biology, physics, poly sci) frequenctly use machine learning algorithms (regression analysis) as part of their normal routine.

Rather than an 'AI' approach, why not investigate some of the more classical problems, such as time series analysis or curve fitting. These are areas where humans and machines continue to butt heads.

only alvaro's answer is useful. Start here for your English assignment.

http://en.wikipedia.org/wiki/Strong_AI#Possible_explanations_for_the_slow_progress_of_AI_research. Follow up on sources.

Required reading (you can probably find it in a library)
http://www.amazon.com/What-Computers-Still-Cant-Artificial/dp/0262540673

The central premise is that "real human AI" is an extremely hard problem mainly because we can't define what that means, but also because there is all this stuff we take for granted. Most attempts at human-level AI are always promising until they hit this giant wall of things we don't typically consider.

Scene
Little girl and mom standing outside a pet store. Girl is pointing towards the window in which 5 puppies are playing with a ball in a pen.
Girl: Mommy! Mommy! I want one!!

Question: how do you know that the girl is talking about a puppy instead of: a ball, a window, the air in front of her finger, a pen, a litter of puppies, a pet store, a pile of poop?

-me

Just pick up any textbook on computer game AI from your library.

Invariably the first chapter is an overview of the history of AI - the failure of classical/symbolic AI and the rise of behaviourism etc.

I say computer game AI because you mention "intelligent by human standereds of an intelligent person". For that you want things that can act in a context that humans will percieve as intelligent. So you want things like perception and adaptivity, rather than just raw number crunching capability. Computer games emphasize things that approximate biological intelligence (i.e. virtual creatures), as opposed to academic AI, which emphasizes boring things like scheduling (the exception perhaps being autonomous robotics).

Look into the Turing Test (basically if a chat-bot can fool a human into thinking they are chatting to another human), and the controversy of whether it is a good measure of intelligence or not (not much of a controversy anymore - generally accepted that it's a poor measure).