However, there are a growing number of teams around the world trying to address the AI evaluation crisis.

MIT researchers developed SEAL, a framework that lets language models continuously learn new knowledge and tasks.