Exploring AI's Self-Improvement Limits in Science and Finance

wordgrammer · x · 2026-07-11

Regarding the feasibility of AI's Recursive Self-Improvement (RSI) in science, the author argues that without human intervention, AI might generate a massive amount of fake scientific progress. Using the stock market as an example, the author notes that while quantitative hedge funds continuously iterate models through backtesting, their returns inevitably slow down. This is due to the rapid decay of market alpha and how large institutions' own trades impact the market. This demonstrates the inherent limitations of purely data-driven self-evolution in complex real-world systems.

Related event: AI Recursive Self-Improvement Concept Faces Heavy Skepticism(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →