From what I read, AI "winters" have not just had less progress but also have had less research activity in general: during them, AI was seen as a dead end that

From what I read, AI "winters" have not just had less progress but also have had less research activity in general: during them, AI was seen as a dead end that didn't provide value, which translates to less funding, which translates to less people doing research on it (or introducing AI to the research they wanted to do already).