Google DeepMind has announced a new framework for measuring progress toward artificial general intelligence, built around cognitive principles. The announcement also includes a Kaggle hackathon intended to develop evaluations that fit within this framework.
The framework is positioned as a way to assess how close AI systems are to AGI in a more structured manner. By inviting the broader community to contribute evaluation methods through the hackathon, DeepMind is seeking input on how such progress should be measured.
Since this is a single announcement, there are no competing perspectives to compare. The blog post focuses on the framework's introduction and the call for evaluations, leaving the technical details of the cognitive framework to be explored further.