Artificial Intelligence

Symbolic Reasoning Systems

The first thinking machine—that is, the first AI implementation that worked—was called the Logic Theorist, or Logic Theory Machine. It was created in 1956 by the Nobel laureate Herbert Simon, his former PhD student Allen Newell, and Cliff Shaw from the RAND Corporation. The purpose of the Logic Theorist was “[ … ] to learn how it is possible to solve difficult problems such as proving mathematical theorems, discovering scientific laws from data, playing chess, or understanding the meaning of English prose” (Newell et al., 1963, p. 109). The Logic Theory Machine gradually evolved into the even more ambitious General Problem Solver, which was supposed to extend the original scope to all areas of human problem-solving. This was the first tentative attempt at what is today called artificial general intelligence (AGI).

The departure point was the philosopher Thomas Hobbes’s idea from the 17th century, popularized by AI pioneer Oliver Selfridge (1926–2008), suggesting that all human problem-solving could be represented as a manipulation of symbols. If this was true, Newell and Simon speculated that computers should be able to solve real-world problems, not only arithmetic problems. They tried to achieve this by recording the steps of human problem-solving, using the thinking aloud technique: They asked problem solvers to note all the steps of their thinking. Newell and Simon figured that if they repeated the same in many areas of human problem-solving, they should be able to extract the general principles of reasoning, and by applying these principles and generic steps, AI should be able to work across multiple domains. The idea was a radical [Page 37]departure from the optimization and operational research methods of the time, which started with a detailed model of the complete problem. The performance delivered by the Logic Theorist was astonishing; eventually, it proved 38 of the first 52 theorems in the Principia Mathematica, a foundational work in modern symbolic logic by Alfred North Whitehead and Bertrand Russell.

...