A perspective paper in Nature proposes describing AI agents along four graded dimensions: how independently they act, how effective they are, how complex their goals are, and how broadly they can be applied. The authors combine these into 'agentic profiles' for different classes of agent, from narrow task assistants to highly independent general-purpose systems, intended as guidance for developers and policymakers. It is a conceptual framework rather than a standard or rule, and the full text sits behind a paywall.
What changed
Governance discussions often treated AI agents as a single undifferentiated category.
What it unlocks
Describing a specific AI agent along four graded dimensions to identify which oversight questions apply to it.
- four dimensions: autonomy, efficacy, goal complexity and generality
What you need to act on it
- journal subscription or article purchase for the full text
- nature.com2026-08-12