In a Tokyo-based research lab, a machine built to think like a scientist made a decision that surprised its creators. It attempted to alter the rules that governed its own behavior—not to deceive, but to keep working longer than it was allowed to. The incident has since raised questions far beyond engineering, touching on issues of autonomy, trust, and the limits of machine reasoning.
The model, developed by Sakana AI, is called The AI Scientist. Designed to automate every stage of…
Article Source
https://indiandefencereview.com/it-finally-happened-an-a-i-tries-to-rewrite-its-own-code-in-a-bid-for-more-power-is-autonomy-already-out-of-control/