The Economics of Recursive Self-Improvement: Is AI Acceleration Inevitable?

The Economics of Recursive Self-Improvement [pdf]

We model the feedback loops driving recursive self-improvement in AI to determine if a self-sustaining acceleration is possible. Our analysis distinguishes between narrow benchmark gains and broad economic value, revealing that current data suggests we are not yet experiencing an intelligence explosion. However, as AI's contribution to research productivity grows, the conditions for rapid, autonomous advancement appear to be strengthening.

"A back-of-the-envelope calculation suggests that feedback loops are not currently strong enough to generate a self-sustaining acceleration, though they appear to be strengthening."