Sakana AI's self-improving model gets better at solving and judging, but can't catch its own blind spots
🤖 AI-generated content — The title and summary were produced automatically by artificial intelligence, without human editorial review.
Researchers from Sakana AI and UC Berkeley have introduced "Multi-Agent Self-Supervision" (MASS), a recursive self-improvement method where a single language model solves tasks, refines workflows, and judges results. While achieving a win rate of nearly 70 percent against the base model after two rounds, shared blind spots across roles can cause the model to mistake errors for progress and reinforce them during training.
Continue in the app — vote & join in ➔Source: The Decoder (AI) · via ahirlevel.hu