What was the AI winter?

The AI winter refers to two long stretches of AI research — roughly the mid-1970s and again the late 1980s into the early 1990s — when funding and interest collapsed after years of hype failed to deliver, and the field went quiet for most of a decade each time.

What caused it

Both times followed the same script. Researchers and funders got excited about early results and made big promises — machine translation, general reasoning, expert systems that could replace human specialists.

Then reality hit: computers weren't fast enough, there wasn't nearly enough data, and the underlying approaches didn't scale the way anyone hoped. Governments and companies pulled their funding, labs shut down, and "artificial intelligence" became almost a dirty phrase in computer science for years.

Why it matters for today's LLM boom

The current wave — ChatGPT, Claude, Gemini, and the trillions of dollars flowing into AI companies — rides on a similar mix of hype and bold promises about what these models will do next.

Some AI researchers openly bring up the AI winters as a warning: if progress slows down or the return on all that investment doesn't show up, funding and enthusiasm could dry up again the same way. That doesn't mean it will happen — today's LLMs already work well enough to be useful right now, unlike some earlier AI hype.

Still, knowing the AI winters happened is a good reason to stay skeptical of confident predictions that this momentum just continues in a straight line forever.