The winters didn't end AI. They ended a particular idea of what AI was supposed to be.

The researchers who stayed through the cold were working on something humbler: not machines that reasoned from first principles, but machines that learned from examples. It was less elegant, harder to explain, and much harder to build. Decades passed before anyone outside the field noticed.

The next module picks up there, with the people who kept going, the idea they were testing, and why it eventually worked.