Artificial intelligence systems are moving toward greater automation in developing new models, raising renewed concerns about recursive self-improvement, according to Axios.
Recursive Self-Improvement, or RSI, refers to AI systems helping create increasingly capable versions of themselves with limited human direction.
Anthropic says AI now performs about 26% of its research and development work, while AI systems assist with 90% of its R&D activities. OpenAI has also reportedly automated much of its experimental model training.
However, experts caution against treating current developments as proof that AI superintelligence is imminent. A recent analysis found existing AI feedback loops have not reached established RSI benchmarks.
The concept dates back decades, including a 1965 warning from mathematician I.J. Good about machines improving themselves. Researchers are now watching emerging RSI experiments closely.
Related Tweet:
"The last invention that man need ever make": The risk of self-improving AI https://t.co/NqHI1sFIie
— Axios (@axios) September 22, 2026
Also Read:
TIPP Staff
