Loading lesson…
The 1980s AI boom ended when expert systems hit a wall and specialized Lisp machines went obsolete.
The mid-1980s AI boom was real. Expert systems saved money at blue-chip firms, and hundreds of startups sold knowledge-engineering tools. Specialized hardware called Lisp machines, sold by Symbolics and LMI, ran the software at premium speeds. The market for AI hardware and software reportedly passed a billion dollars.
Then it collapsed. Around 1987 to 1988, cheaper general-purpose workstations from Sun and later PCs running LISP matched Lisp machine performance at a fraction of the cost. The specialized hardware industry died within a few years.
By the early 1990s, the term AI carried enough stigma that many researchers rebranded. Machine learning, intelligent agents, and informatics all emerged partly as escape hatches from a tainted word.
By the late 1980s, the business press was speaking of an AI winter.
— Nils Nilsson, AI historian
The big idea: hand-crafted knowledge does not scale. The next era of AI would have to learn from data, because humans could not write enough rules fast enough.
15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-history-second-winter-builders
What technology emerged in the mid-1980s that could perform specialized tasks by following hand-written rules?
Which company sold specialized hardware designed to run Lisp programs faster than regular computers?
What caused the Lisp machine industry to collapse around 1987-1988?
Why were expert systems expensive to maintain over time?
What was the Fifth Generation project?
What was the main limitation of expert systems that prevented them from improving on their own?
By the early 1990s, why did many AI researchers change the names of their work?
Which company was mentioned as a competitor to Symbolics in the Lisp machine market?
What happened to expert systems when the domain they covered changed?
What term did the business press use by the late 1980s to describe the state of the AI industry?
What type of tasks were expert systems designed to handle?
Why did companies quietly retire many expert system deployments?
What programming language was primarily used to build expert systems?
Which factor did NOT contribute to the expert systems collapse?
What new fields emerged partly as 'escape hatches' from the tainted term 'AI'?