1987–1993
The second AI winter
The market for commercial symbolic AI collapses, first LISP machines and then expert systems.
What happened
In 1987 the market for LISP machines, expensive computers built to run that language, collapses within a single year: workstations from Sun and desktop computers from Apple and IBM do the same job for less money. Late that year DARPA abandons the Autonomous Land Vehicle, the flagship project of its Strategic Computing Initiative, and cuts AI funding deeply. By 1989 Digital Equipment had 59 people working just to maintain its rule base. In March 1992 the ten-year term of Japan's Fifth Generation project, launched in 1982, comes to an end; it had cost about $400 million without reaching its goals.
Why
The boom of the 1980s rested on expert systems, and they proved brittle outside their domain and costly to maintain. Most companies never managed to build the team needed to use them. Haigh describes it as a funding bubble that burst.
What kept going
The winter hit symbolic AI above all. By the mid-1980s the work of John Hopfield, David Rumelhart, and others had revived interest in neural networks, and that line of research carried on under other names, such as "machine learning" or "pattern recognition." Speech recognition advanced with hidden Markov models. Many AI methods stayed in use inside larger systems, no longer under that label.
Why these dates
The start, 1987, is set by the collapse of the LISP machine market, and the sources agree on it: Haigh writes that the end of the 1980s is unanimously seen as the beginning of the winter. The end has no date. The 1987–1993 dating is the one Wikipedia gave when Haigh cited it in 2023. That page now extends it to 2000, and Haigh speaks of a slump lasting two decades. That is why the band is drawn solid until 1993 and fades out by 2000.
Events that define it
Sources
Scholarship and context
- How the AI Boom Went Bust. Communications of the ACM, vol. 67, no. 2, pp. 22-26 (opens in a new tab). Thomas Haigh, 2024
- There Was No ‘First AI Winter’. Communications of the ACM, vol. 66, no. 12, pp. 35-39 (opens in a new tab). Thomas Haigh, 2023
- Fifth generation project lives on. Nature, 357, p. 619 (June 25, 1992) (opens in a new tab). 1992