Sakana AI Unveils MASS: A Self-Improving AI Loop Built On Its Own Collective Intelligence
This story is only covered by news sources that have yet to be evaluated by the independent media monitoring agencies we use to assess the quality and reliability of news outlets on our platform. Learn more here.5 Articles
5 Articles
The Sakana AI and UC Berkeley MASS method enhances the performance of a model in several tests and improves its evaluation capacity. However, researchers warn that the same blind spot in the system can make errors or manipulated results appear to be advances and incorporated into subsequent training cycles.
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 lets one language model improve itself by solving tasks, judging results, and refining workflows, reaching a win rate of nearly 70 percent against the base model after two rounds. But when the model misses the same errors across all three roles, it can mistake them for progress and reinforce them during training.
Sakana AI can improve a single language model itself by acting as a problem solver, evaluator and optimizer at the same time. The so-called MASS method increases the victory rate against the basic model to almost 70 percent, but the study also shows a risk: If the model fails in all three roles, it is taken as a step forward into the next training.The article Self-Improving AI shows progress, but flawed results easily pass as success appeared fi…
Coverage Details
Bias Distribution
- There is no tracked Bias information for the sources covering this story.
Factuality
To view factuality data please Upgrade to Premium
