The Logic Theorist: The World's First AI Program

In 1955, Allen Newell and Herbert Simon set out to prove that a computer program could do something considered uniquely human: mathematical reasoning. Their Logic Theorist, developed with Cliff Shaw, was a program that could prove theorems in propositional logic by searching through possible inference rules, much like a mathematician works through a proof. It used heuristic search: not exhaustive brute force, but guided exploration based on what seemed promising.

The Logic Theorist proved 38 of the 52 theorems in Alfred North Whitehead and Bertrand Russell's "Principia Mathematica," the foundational work of modern mathematical logic, published in 1910. One of its proofs was actually more elegant than the original Russell proof. Newell and Simon, delighted, tried to add the Logic Theorist as a co-author on the paper they submitted to the Journal of Symbolic Logic. The journal rejected this arrangement. The program's proof appeared in a footnote.

Newell and Simon brought the Logic Theorist to the Dartmouth Summer Research Project in 1956. It was, arguably, the first AI program to be demonstrated, a working system that did non-trivial intellectual work. Simon reportedly told his students at Carnegie Institute of Technology that Christmas 1955 that "we have invented a thinking machine." The claim was, by any measure, premature. But the existence of a program that could reason about mathematics, not just calculate, was a genuine milestone.

The Logic Theorist was followed by the General Problem Solver (GPS) in 1957, which used means-ends analysis, comparing the current state to the goal state and finding operators to reduce the difference. These programs pioneered symbolic AI: the approach of representing knowledge as symbols and reasoning by manipulating them. This dominated AI research for 30 years, until neural networks displaced it in the late 1980s. But the question Newell and Simon asked, can machines reason?, has never fully been answered.