The “AI winter” was a period of reduced interest in the field of artificial intelligence lasting from the 1980s through the 2000s. This lack of interest led to a lack of funding.
The term was coined by analogy to the hypothesized concept of “nuclear winter” where the fallout from a nuclear war would change weather patterns.
The “winter” has been blamed on several causes: the overhyping of AI research, the interdisciplinary nature of AI and conflicts among university departments, university budget cuts, lack of practical applications for AI research, and cheaper computing products overtaking expensive Lisp machines in performance.
The AI winter of the 1980s was precipitated by failures of “expert systems,” systems that were purported to have decision-making capabilities similar to a human expert. These systems were popular in the 1980s, but they proved expensive and unreliable. This led to the belief that AI research had been overhyped.