Glossary » artificial intelligence
“AI”; a field of study which examines how to perform high-level thinking on computers. Artificial intelligence research is typically associated with domains such as speech synthesis and recognition, language translation, image recognition, and strategy and planning. AI research is also associated with certain types of algorithms (such as inference engines), data representations (such as blackboards), and general models of thought (such as connectionist networks).
Usability issues are critical in many AI systems, where a human works with the system to work out and apply results (as in language translation systems and expert systems), and when the AI system serves as the user interface for the user (as with speech systems). AI also is applied in some systems to build a computer model of the user, which is then used to help anticipate the user’s needs and optimize the interface (as in computer-aided instruction systems and adaptive systems).
AI is part of the field of cognitive science. Approaches to AI typically seek to build systems that can act intelligently in one way or another, often taking approaches modeled on human cognition, but not constrained to a human model. This contrasts with the field of cognitive psychology, which seeks to create computational models of human cognition specifically to build a more complete understanding of human cognition.
The term “artificial intelligence” is misleading to the extent that, to begin with, we really don’t have a good grasp of what “intelligence” is. Human intelligence is often considered to be characterized by those properties of thought that are not demonstrated by other animals, such as language, long-term planning, symbolic manipulation, reasoning, and meta-cognition. However, arithmetic, for example, which is unique to humans, is nevertheless not considered to be the proper domain of artificial intelligence.
The most common criterion considered for evaluating whether a computer has achieved human intelligence is the Turing test. In the Turing test, a person communicates via a text terminal with two hidden conversational partners: another human and a computer. If the person cannot distinguish between the human and the computer, then the computer would be considered to be behaving intelligently. However, numerous problems with this definition of intelligence exist. For example, in practice, when this test is applied, humans are often mistaken for computers, indicating that we don’t really have a good sense of what we mean by considering a human to be intelligent in the first place!