Morris, M. R., Sohl-Dickstein, J., Fiedel, N., Warkentin, T., Dafoe, A., Faust, A., … & Legg, S. (2023). Levels of AGI: Operationalizing Progress on the Path to AGI. arXiv preprint arXiv:2311.02462.
This paper introduces a framework to classify the capabilities and behavior of Artificial General Intelligence (AGI) models. It proposes “Levels of AGI” based on performance and generality, aiming to offer a common language for comparing models, assessing risks, and tracking progress toward AGI. The framework focuses on capabilities rather than processes, and emphasizes the need for benchmarks to quantify AGI attributes. It also discusses the interaction of AGI levels with deployment considerations like autonomy and risk, highlighting the importance of responsible Human-AI Interaction paradigms.