Why did AI funding collapse in the 1970s and again in the late 1980s

Two Separate Collapses, One Recurring Pattern

AI research funding collapsed significantly on two separate occasions — in the 1970s and again in the late 1980s and early 1990s — and while the specific triggers differed each time, both episodes shared an underlying pattern: a widening gap between promised capability and what the field could actually deliver in practice.

The 1970s Collapse: Overpromising and Critical Reports

The first collapse, discussed in more detail elsewhere, stemmed largely from overly ambitious early predictions about how quickly AI research would achieve major milestones, combined with influential critical government reports — including the UK’s Lighthill Report and the earlier US ALPAC report on machine translation — that led major funders to substantially cut research support after those ambitious timelines weren’t met.

The Late-1980s Collapse: The Rise and Fall of Expert Systems

The second collapse followed a different pattern, tied closely to the commercial rise and subsequent disappointment of “expert systems” — AI programs designed to capture and apply the specific, codified knowledge of human experts in fields like medical diagnosis or industrial troubleshooting. These systems had generated substantial commercial enthusiasm and investment in the early-to-mid 1980s.

Why Expert Systems Fell Out of Favor

In practice, many organizations found that expert systems were considerably more expensive to build and, critically, to maintain than initially expected, since keeping the encoded expert knowledge current and comprehensive required ongoing, labor-intensive effort. Many systems also proved brittle and difficult to scale beyond their originally narrow, well-defined problem domains, disappointing commercial adopters who had expected broader, more flexible value.

The Parallel Collapse of Specialized AI Hardware

Compounding this disappointment, a thriving commercial market for specialized computer hardware optimized specifically for AI software — known as Lisp machines — collapsed by the late 1980s, as increasingly powerful and more affordable general-purpose computers became capable of running similar software, undercutting the specialized hardware companies’ business model and contributing to broader investor pullback from AI-focused ventures.

Why Understanding Both Episodes Matters

Both funding collapses illustrate a recurring dynamic in AI’s history: periods of intense optimism and investment, followed by disappointment when practical results failed to match ambitious promises, leading to sharp funding contractions — a pattern worth keeping in mind when evaluating the sustainability of the field’s current, more recent boom.

Bottom Line

AI funding collapsed in the 1970s due to overpromised results and critical government reports, and again in the late 1980s and early 1990s due to the commercial disappointment and high maintenance costs of expert systems combined with the collapse of the specialized Lisp machine hardware market — two distinct episodes united by the same underlying pattern of promised capability outrunning delivered results.

Look Up AI Terms

Search plain-English definitions of AI and machine learning terms in our free AI Glossary.

Go deeper

• How do search and rescue robots navigate through unstable or collapsed structures?