🤖 AI Summary
The absence of a unified evaluation framework hinders rigorous assessment of progress toward Artificial General Intelligence (AGI).
Method: This paper introduces the first three-dimensional AGI capability taxonomy—spanning performance depth, functional breadth, and autonomy—grounded in cross-literature conceptual analysis and human-AI interaction safety requirements. We construct an AGI ontology satisfying six criteria: comparability, measurability, forward-lookingness, interpretability, scalability, and safety-awareness. Crucially, we extend traditional unidimensional performance-centric grading by formally integrating generalization capacity and autonomy as core grading principles.
Contribution/Results: The framework enables hierarchical positioning of pre-AGI models for the first time, supporting cross-model capability benchmarking, risk-tiered evaluation, and design of trustworthy, autonomy-aware benchmarks. It provides an actionable, graded autonomy framework to inform AGI development metrics and safe deployment strategies.
📝 Abstract
We propose a framework for classifying the capabilities and behavior of Artificial General Intelligence (AGI) models and their precursors. This framework introduces levels of AGI performance, generality, and autonomy, providing a common language to compare models, assess risks, and measure progress along the path to AGI. To develop our framework, we analyze existing definitions of AGI, and distill six principles that a useful ontology for AGI should satisfy. With these principles in mind, we propose"Levels of AGI"based on depth (performance) and breadth (generality) of capabilities, and reflect on how current systems fit into this ontology. We discuss the challenging requirements for future benchmarks that quantify the behavior and capabilities of AGI models against these levels. Finally, we discuss how these levels of AGI interact with deployment considerations such as autonomy and risk, and emphasize the importance of carefully selecting Human-AI Interaction paradigms for responsible and safe deployment of highly capable AI systems.
