A computer passes the Turing intelligence test for the first time

In 1950 Alan Turing proposed what would later be known as the Turing test in a paper he published. This paper proposed that a computer must display enough intelligence to convince a jury that it is, in fact, human. This can be tested by having a conversation through the computer with the judges at the competition. At this point, minimum 30% of the judges have to be convinced that behind the computer there might actually be a real person.

Who was Alan Turing?

Alan Turing was a British mathematician that is considered one of the fathers of modern computing and informatics. During World War II, he created at Bletchley Park the first computer that could decipher the German codes. This helped to win the war in the end.

The computer was called Enigma:

enigma computer alan turing

Image source: Wikipedia

Who managed to beat this proposed test?

The program that beat the Turing test is called Eugene Goostman and is, in essence, a chatterbot. It was created in 2001 by two scientists at the university of St. Petersburg and it’s designed to act like a 13 year old boy. In 2012, this program, which is not artificial intelligence (for those of you wondering), managed to convince 29% of the judges, but that was not considered good enough since the minimum requirement is 30%. Well, on the 7th of June, 2014, this program competed again in a competition held by the Reading University for the Turing test. It managed to convince 33% of judges that behind the computer may be a real person. Of course, this was helped by the fact that the program was designed to act like a 13 year old boy, that is not expected to know many answers, or answer very difficult questions. Te website where the robot was hosted is: http://www.princetonai.com/bot, but it seems to be down at the moment. I do remember that a long time ago, I played with another chatterbot called the Alicebot.

Why this isn’t artificial intelligence?

For something to be considered true artificial intelligence, it would have to do some things that only humans have been able to perform like:

  • lingvistic perception
  • visual perception
  • decision abilities
  • recognizing it’s own existence
  • the ability to learn on it’s own

In fact, this program is based on algorithms and dictionaries that help it process the questions it gets asked and return an answer that would be as real as any real person would. Although this isn’t artificial intelligence, it does put us a small step closer to that goal.

Tell me your thoughts...