A. I.

(Don't be lazy, I write these in english for YOUR own education: read 'em.)

Artificial Inteligence is the study that overlaps fields such as computer science, psychology, engineering, neurology and even philosophy; aiming to create machines with intelligent behavior, learning and adaptation.

Whenever we hear about A.I. we usually think about futuristic anthropomorphic robots such as the ones depicted in works such as I Robot or A. I., or smart military supercomputers such as Terminator movies' Skynet.

Nevertheless, interesting A. I. projects currently exist within the reach of common people, and this post means to review one of them: 20Q.

20Q consists in a game where the user thinks of an object, and then answers questions posed by the program in a "Yes" or "No" answer fashion, which lead the program to the correct guess. "Unknown", "Irrelevant", "Sometimes","Doubtfully" and "Probably" answers were later added. The game's objective is to force the program to ask more than 20 questions before guessing what you're thinking. You'd think it's quite easy to accomplish, and in the program's early stages it may have been so, but I assure you, the program is astoundingly efficient at the present time.

20Q was created in 1988 by Robin Burgener as an A.I. experiment, based on the classic game Twenty Questions. This program learns only from interaction with users, it has no programmed answers. 20q uses an artificial neural network to pick the questions and then make the proper guesses.

There are versions of 20Q in 19 different languages, and there are also specialty variations of the program's database, such as "music" or "movies". 20Q has a success rate between 73 and 78%. The autor says this is because the algorithm is intentionally "flawed", so that the user has more chances to beat the program, keeping it entertaining and therefore allowing the program to continue learning.

Try it out @ http://www.20q.net