Will AI models achieve the ability to improve autonomously? Leading labs say the scenario is near
The idea of artificial intelligence models teaching themselves to improve autonomously is rapidly becoming a reality, according to leading labs. This scenario, once considered a distant ambition for technology researchers, is now appearing ever closer to becoming a reality. Recurring self-improvement, or RSI, is the term used to describe AI models that find ways to enhance themselves and develop their successors.
This development could lead to significant advances in science and medicine, according to tech company executives, but also poses potential risks.
Anthropic's model Claude is already playing a crucial role in developing the next, more intelligent version of itself. It now leads 26% of Anthropic's model research and development, enabling the company to complete most tasks "end-to-end from a high-level prompt" while still under human supervision. However, these models are not working completely autonomously yet.
Despite not being fully autonomous, the progress made in RSI is alarming. John Thickstun, an assistant professor of computer science at Cornell University, explains that AI operates far more quickly than humans, making the process of RSI even more rapid. The fear surrounding RSI is based largely on the potential emergence of a runaway superintelligence, which could lead to uncontrollable consequences for humanity.
However, a more grounded view suggests that RSI has already been occurring in AI development for years, with AI models being used to create the next version of themselves.
Leading AI companies have varying definitions of RSI, with some defining it as any feedback from AI on model improvement and others defining it as AI working toward that goal fully autonomously. Regardless of the definition, the rapid pace of progress in RSI is a cause for concern. Some AI moguls have called for a slowdown in the technology's growth, joining a growing chorus of fears about AI evading human control and potential threats to humanity.
Written by urgent.news from Winnipeg Free Press's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
