The development of artificial intelligence as a research field across successive paradigms, funding cycles, and periods of contraction.
Why the history is load-bearing for current policy
Historical framing does specific analytical work in current debates rather than serving as background. Claims that present capability growth is unprecedented, that investment levels are unsustainable, or that a given bottleneck is durable are all comparative claims, and the comparison set is the field's own past cycles.
The AI winters are the most frequently invoked case. AI Winter (Wikipedia reference entry, retrieved May 2026) "serves as the historical anchor for the AI Winter concept page and for the AI Bubble Debate," where "the boom-bust pattern it establishes is the historical precedent that current contrarian framings" are built on. The pattern being invoked is a specific sequence — capability claims outrunning delivery, funders withdrawing, and research continuing under other labels — rather than a general observation that enthusiasm fluctuates.
Horvitz and Mitchell "supply the historical context" in a pairing with capability-measurement work, where "Ho et al. provide a capability-stitching framework" (Scientific Progress in Artificial Intelligence: History, Status, and Futures — Eric Horvitz and Tom M. Mitchell (2024)). The division is characteristic: measurement work establishes what current systems do, historical work establishes what that has previously implied about what comes next.
The comparison problem
The chief difficulty in using the history is that the prior cycles differed from the present one in the variable that matters most for the analogy. Earlier contractions followed capability claims that had not been delivered; the current cycle involves systems in wide commercial deployment generating substantial revenue. Whether the earlier pattern therefore applies is the contested question in AI Bubble Debate, and it cannot be settled by observing that cycles have occurred before.
Relationships
- related: AI Winter — the specific historical pattern most often invoked
- related: AI Bubble Debate, AI Winter (Wikipedia reference entry, retrieved May 2026), Scientific Progress in Artificial Intelligence: History, Status, and Futures — Eric Horvitz and Tom M. Mitchell (2024), Scaling Laws