An AI winter is a period when confidence, investment, and research funding in artificial intelligence decline sharply because AI systems have failed to meet inflated expectations. It does not mean that all AI research stops; progress continues, but with less public attention and financial support.
Why AI Winters Happen
AI development often follows a cycle:
- A technical breakthrough produces ambitious predictions.
- Governments, universities, and companies increase funding.
- Existing systems encounter limitations, such as insufficient computing power, poor-quality data, narrow capabilities, or high costs.
- Promised applications fail to appear quickly enough.
- Investors and funding bodies reduce support, creating an AI winter.
This connects to the evolution of AI, syllabus area 3.6D, which is shared by SL and HL students. It also illustrates change because technological development involves periods of acceleration, slowdown, continuity, and renewed growth rather than automatic progress.
| Period | Main explanation |
|---|---|
| First AI winter, mainly the 1970s | Early AI could solve restricted problems but struggled with complex real-world situations. Critical evaluations and unmet predictions contributed to funding reductions. |
| Second AI winter, mainly the late 1980s and early 1990s | Commercial expert systems were expensive to maintain and often too inflexible, while the specialized hardware market collapsed. |
System: Expert systems used encoded rules to reproduce decision-making in narrow fields. Example specifics: During the 1980s, companies invested heavily in these systems, but many proved costly and difficult to update. Impacts: Some organizations reduced AI investment, and specialist companies failed. Implications: The decline showed that enthusiasm and funding depend on social and economic expectations, not only technical performance.
A common misconception is that an AI winter means AI became completely inactive. In reality, research continued and later advances in computing power, data availability, and machine learning helped renew interest.