Sakana AI's self-improving model gets better at solving and judging, but can't catch its own blind spots

Sakana AI's MASS method, which lets a single language model improve itself by solving tasks, judging results, and refining workflows, reached a win rate of nearly 70 percent against the base model after two rounds, The Decoder reported. The method fails when the model makes the same errors across all three roles, causing it to mistake those errors for progress and reinforce them during training.