The term 'Artificial Intelligence', or 'AI', refers to a branch in computer science that deals with the simulation of human intelligence in computers.   The term 'artificial intelligence' was coined in 1956 by John McCarthy, who defines it as "the science and engineering of making intelligent machines."  The pioneers in the field had always been optimists who believed that intelligent machines will be here sooner than later, but successive setbacks in the field have proven time and again that, as far as machines are concerned, 'common sense' isn't common after all.

Even the exact definition of what an 'intelligent machine' is remains unclear.  Is a robot that's capable of making unique portraits of people using its own stored rules for creative drawing already considered 'intelligent'  (such a robot already exists, by the way), or does it take more than that?  Many AI experts believe that an intelligent agent (that's what they call a system that has artificial intelligence) should be able to perceive its environment and adapt to it so that it can take actions that maximize its chances of successfully completing its task.  The ability to adapt to an unexpected situation is therefore a big component of artificial intelligence.

AI research as a field started in 1956, when it was founded by AI pioneers that included McCarthy, Marvin Minsky, Allen Newell, and Herbert Simon in a conference in Dartmouth College.  These pioneers and their students wrote programs that astonished many people at that time, including one that speaks English (or appears to) and another that could solve word problems in algebra.  Because of such breakthrough programs, AI research got heavy funding in the 1960's, especially from the U.S. Department of Defense.

    

Figure 1.  The big AI question: Will this robotic Einstein be as intelligent as the real Einstein someday?

            

Before long, AI scientists came face to face with the reality that simulating the intelligence of the human brain is not easy. Subsequent AI projects simply failed to demonstrate that machines can be as smart as people, causing the funding for AI research to dwindle in the 1970's.  In the 1980's, however, things started to look bright once again for AI researchers, thanks to the emergence of special computer programs called 'experts systems'. An expert system is a computer program designed to solve specific real-world problems in the same way that a human expert would.  Expert systems became commercially successful by specializing in narrow fields, which enabled them to provide real and practical solutions to special problems.

  

By the late 80's, the future of AI became questionable again. A subsequent dry spell in funding forced AI to be pursued more prudently.  Still, the 1990's and this century produced significant progress in AI research, helped in part by the rapid advancement of the power of computers.

  

The science of artificial intelligence is so complex that it has diverged into many sub-fields of specialization. Problems being addressed by artificial intelligence research today include:  problem solving and planning techniques, knowledge representation, learning, natural language processing, motion and object manipulation, vision and perception, creativity, social intelligence, and general intelligence.

    

   

See Also:   Artificial Neural Networks,  Natural Language Understanding,  More Articles