From Seed AI to Technological Singularity via Recursively Self-Improving Software
This foundational work aims to inform researchers and policymakers about the potential risks and dynamics of self-improving AI systems, though it is incremental as it builds on existing literature.
The paper tackles the problem of understanding and predicting the behavior of recursively self-improving software, introducing RSI Convergence Theory to analyze its general behavior and addressing security implications.
Software capable of improving itself has been a dream of computer scientists since the inception of the field. In this work we provide definitions for Recursively Self-Improving software, survey different types of self-improving software, review the relevant literature, analyze limits on computation restricting recursive self-improvement and introduce RSI Convergence Theory which aims to predict general behavior of RSI systems. Finally, we address security implications from self-improving intelligent software.