What is Artificial Intelligence?

The eventual goal of artificial intelligence research is to make thinking computers. The precise definition of “thinking” is debatable, and some philosophers contend that a machine is principally incapable of adequately emulating the mental life of a human being. However, there is no doubt that computers can mimic some human behavior. In fact, Alan Turing proposed a test for determining whether a computer can be considered a “thinking machine”: the computer is placed behind one curtain, and a person is placed behind another one. An interviewer, who cannot see behind the curtains, asks questions of both subjects. Based on the answers, the interviewer tries to determine which one is the machine. A computer that perfectly imitates a human being would be able to deceive the interviewer half the time.

Recent Developments in Artificial Intelligence

Turing predicted that by the year 2000, computers would be sophisticated enough that an interrogator would not be able to identify the computer correctly more than 70 per cent of the time after a five minute conversation. This prediction turned out to have been somewhat optimistic, but substantial progress has been made in many areas. Some recent achievements include:

  • Speech dialog systems by firms like IBM, Dragon and Lernout & Hauspie have introduced voice interfaces to the consumer market.
  • Expert systems and case-based reasoning have been deployed to solve practical problems: a computerized leukemia diagnosis system did a better job checking for blood disorders than human experts.
  • Machine translation has become more widespread, with such services as Babelfish offering automatic translations between many languages.
  • Deep Blue was the first computer to beat a human chess Grandmaster, Gary Kasparov.
  • Fuzzy controllers in dishwashers and other appliances

Each of these results is a step towards an artificial human being, and the pace of progress is quickening. Ambitious experiments are underway that aim to create complex, humanoid robots. A small company called IS Robotics has created a cheap prototype of an emotive robot, named IT, which expresses loneliness, amazement, joy, and other human emotions. Even though the underlying software may not be nearly as complex as the human brain, IT is believable enough to evoke emotions in humans, so it can attract attention for marketing purposes. "If you went up to a whimpering robot in a store, and it suddenly stopped crying and got happy that you were there, you'd get a warm and fuzzy feeling," explains Colin Angle, the company’s president.

At the MIT Artificial Intelligence Laboratory, the Cog Shop seeks to “bring together each of the many subfields of Artificial Intelligence into one unified, coherent, functional whole.” There, scientists are building Cog—a robot that will eventually have vision, hearing, touch, and humanoid form. The hypothesis behind Cog is that a human intellect is at least in part formed by human interactions, which include physical activity. By modeling these interactions, Professor Rodney Brooks, the principal investigator, hopes to test this hypothesis and to learn more about the constitution of human thought.

To complement this hardware-oriented approach, several research groups are also tackling the problem of artificial intelligence from a software viewpoint. The Formal Reasoning Group and Knowledge Systems Laboratory at Stanford, for instance, focus on simulating the acquisition, processing, and application of information. These techniques need not strive towards modeling a human conscience; they can also serve to increase the computational power of algorithms that are unique to machines. In other words, artificial intelligence could turn out to have very little similarity to humans.