The ultimate aim of artificial intelligence is to create a machine that can think better than any human, and this scares a lot of people. Many movies have been made to cash in on the widespread fear of computers - the idea that something more intelligent than humans would want to destroy or subjugate the human race is a common one.

Probably the first popular scary movie about artificial intelligence would have been 2001: A Space Odyssey by Arthur C. Clarke. It told the story of several astronauts in a futuristic space ship on a mission to Jupiter. Their ship was controlled by an extremely advanced computer known as HAL (an acronym for Heuristic Algorithmic Logic), which was capable of advanced tasks such as speech and chess-playing. Chess was chosen here due to its obvious difficulty - HAL's ability to win games against thinking humans was thought a near-impossible task at the time, and it showed how impressively advanced he was.

Amusingly enough, the story is set in the year 1991; at the time it was written, people expected that thinking, talking, chess-playing computers would exist by then. So far computers can talk, although they still can't "think" as such. The only one of HAL's capabilities that has been comprehensively programmed so far is the chess-playing ability: IBM's famous Deep Blue system. Chess is a particularly annoying computer problem as playing well requires a huge amount of knowledge and intuition on the part of the player. This cannot always be simulated by a computer, but as processor speeds increase, the analytical capabilities of computers are increased accordingly.

There is a limited number of first possible moves in the game. With only eight pawns to move at the start, and a correspondingly limited number of second moves, considering that only other pawns or pieces uncovered by the pawn movement can be moved, it is possible to store a database of popular openings. This is why computer chess programs often take a very short time to respond to your first few moves in a game, but suddenly slow down later.

Once past the first few moves, the analytical power of the chess program is needed. The commonly accepted way to win chess games is to look as far ahead into the future as possible, guessing what your opponent will move and selecting the right pieces to move in order to combat your opponent's plan. Computers are suited to this, but after several moves, the number of possible future moves to predict quickly becomes unmanageable on even the most powerful computer systems around. The art of successfully designing a good chess-playing computer is to have a form of hard-wired or learning artificial intelligence past the pure algorithmic nature of the game.

Processor power matters too, and that's where IBM has its strength: the miraculous Deep Blue chess computer, which beat world champion Gary Kasparov several months ago in the rematch between the two. Kasparov beat the computer in their first match, but the updated computer triumphed in the rematch. IBM's computer was an RS/6000, a massively parallel form of computer, designed to handle huge amounts of processing. For example, IBM's web servers are run on RS/6000 computers, and IBM sells mail servers that can handle 500,000 users, upgradeable (by plugging in more parallel processors) to 1,000,000, 1,500,000, or even more. So this is what it takes to beat a grandmaster :-)

The Terminator and Terminator II: Judgement Day paint a picture of a dark future, in which the only surviving humans are part of a dwindling resistance movement against an army of artificial life. The scenario is that a neural network named "SkyNet" was placed in charge of defending the world (from unnamed invaders, or perhaps from the world itself) and it "turned bad" after being active for a period of time. SkyNet's ability to improve itself and adapt to new situations, characteristic of neural networks and some other AI methods, allowed it to make the decision to declare war on the rest of the world, with spectacular effects.

A popular role-playing game (in the same vein as Dungeons & Dragons and many others) named Paranoia is set around the year 2300. The world has been destroyed by a nuclear attack, triggered accidentally by a planetoid on a collision course with the Earth, and all of the world's population is living in a collection of domed complexes, each run by giant computer networks. The games take place inside Alpha Complex, the dome over San Francisco, which has isolated itself from the rest of the world and now lives in paranoid fear of invasion, its programs and systems slowly degenerating. Alpha Complex's network probably ran a hybrid AI system, likely a cross between neural networks, a system of genetic algorithms, and expert systems.