From general reasoning to expert knowledge
Symbolic AI starts out looking for a general method of reasoning and ends up finding its strength in the specialized knowledge of expert systems.
Logic Theorist (1955)
Newell, Shaw, and Simon write the Logic Theorist, which proves theorems by searching a tree of possibilities guided by heuristics. To write it they create IPL, a list-processing language.
General Problem Solver (1957)
The same authors generalize the search to any problem that can be formalized, using means-ends analysis, kept separate from the knowledge specific to each problem.
LISP (1958)
McCarthy takes IPL's list processing and creates LISP, which will be the language of AI in the United States for three decades.
DENDRAL (1965)
DENDRAL, written in LISP, infers chemical structures using chemists' knowledge. Its lesson is that the power lies in knowledge, not in a general method.
STRIPS (December 1971)
STRIPS plans the actions of the robot Shakey by searching, GPS-style, for a sequence that narrows the gap between the current state and the goal.
MYCIN (1972)
MYCIN brings DENDRAL's approach to medicine, with some 600 rules about infections and antibiotics.