Tuesday, February 15, 2011

Man and machine are draw at Jeopardy!

Draw: 5000 dollars for the machine, 5000 and 2000 dollars for two people who attended the first day of the challenge between human and artificial intelligence in the game show Jeopardy!. This is a classic American television: the presenter read a reply, and competitors must guess the question pertaining thereto.

In a first Test match, the computer had dominated the early game, before falling behind by one of the human players. Designed by IBM, Watson, artificial intelligence, which participates in the game on three days (14, 15 and 16 February), showed during the first part he could suffer from weaknesses very "human".

The computer has repeated a false answer that had already been given by one of his opponents, he has also shown that in the finals, where players bet on the reliability of their answers, it tends to show somewhat confident, leaving a window that its opponents, the game show veterans, were able to exploit.

Watson had yet important theoretical advantages over human players. Besides access to a database of general knowledge, useful in determining the correct answers, the machine can, unlike human players, press the buzzer on instantly and take advantage of a crucial advance. SOME GAMES ARE HUMANS Dominated by The experiment conducted by IBM of particular interest when compared to tests conducted on other games.

In chess, the computer excels thanks to its ability to calculate a huge number of possibilities, without being able to reproduce the talent of the best human players to focus on promising avenues of play. But Jeopardy! uses any other skills, including language comprehension. One area where artificial intelligence still maintains a strong margin growth in machine translation as in speech recognition.

But despite the tremendous increase in processing power, and thus the computing capacity of the machines, some games among the "mathematics" are still under the domination of the best human players. The game of go, with its countless possible combinations, for example, is too complex for even the most powerful computers, while human players know how to determine the best shots without having to calculate one by one the consequences of all possible choices.

No comments:

Post a Comment