AI Winter describes the cyclical downturns in the field of artificial intelligence when the high expectations set by initial breakthroughs are not met, leading to diminished enthusiasm and investment. During these periods, AI research faces significant setbacks as public and private sector funding dries up, media coverage wanes, and academic interest shifts elsewhere. These downturns often follow periods of intense hype and promise, where the limitations of current technologies become apparent, and the gap between ambition and practical application leads to disillusionment. AI Winters have significantly impacted the pace of AI development, causing delays in advancements and shifting the focus towards more feasible projects.
The term "AI Winter" first emerged in the late 1970s and early 1980s, with the first major AI Winter occurring around 1974-1980 due to the limitations of expert systems and other early AI technologies. Another notable AI Winter occurred in the late 1980s and early 1990s when the promise of fifth-generation computer systems did not materialize. These periods became more widely recognized as distinct phenomena by the mid-1980s and again in the early 1990s.
Significant figures in identifying and analyzing the phenomenon of AI Winter include Marvin Minsky and John McCarthy, pioneers in AI who also critically evaluated the realistic progress and limitations of AI research. Additionally, the UK government's Lighthill Report (1973), authored by Sir James Lighthill, played a pivotal role in triggering the first AI Winter by highlighting the lack of practical applications of AI research at that time.