Symbolic AI
The classical AI paradigm based on logic, rules, and explicit knowledge representation. Dominated from the 1950s through the 1980s.
Sub-topics
John Haugeland's term for classical symbolic AI: intelligence as formal symbol manipulation. Dominated the field from the 1950s to the 1980s.
Programming paradigm based on formal logic. Prolog (1972) became the primary language. Japan's Fifth Generation Computer project (1982) was built on it.
Rule-based systems encoding domain expert knowledge. MYCIN (1976) diagnosed infections; XCON (1980) configured DEC computers. Peak popularity in the 1980s.
Methods for encoding information about the world in forms that AI systems can reason about: semantic networks, ontologies, description logics.
AI systems that generate sequences of actions to achieve goals. STRIPS (1971) introduced the classical planning formalism still used today.
Early natural language processing using grammars, parse trees, and hand-crafted rules. Chomsky's formal grammars and Winograd's SHRDLU (1970) exemplified this approach.
Formal specifications of shared conceptualizations. Cyc (1984) attempted to encode all common sense knowledge. OWL became the standard for the Semantic Web.
Tim Berners-Lee's vision (2001) of a machine-readable web using RDF, OWL, and SPARQL. A modern evolution of symbolic knowledge representation ideas.
Using formal logic to automatically prove theorems and verify programs. Resolution theorem proving (1965) and SAT solvers underpin modern verification tools.
Minsky's 1974 frame theory: knowledge as stereotyped situations with default values. Influenced object-oriented programming and modern knowledge graphs.