levels of autonomy
In plain wordsLadders that grade how independently an AI system acts — from a tool a person fully controls, through suggesting and asking for approval, up to acting on its own while people merely watch.
Graded scales that classify AI systems by how much they act without human involvement, used to set expectations, design oversight and scope rules. The template comes from human-factors engineering: Thomas Sheridan and William Verplank's ten-level scale of computer autonomy (1978), refined by Raja Parasuraman, Sheridan and Christopher Wickens (2000), and SAE International's six levels of driving automationlevels of driving automation, n.: A six-step scale, from Level 0 to Level 5, describing how much of the driving a car's automation does and how much the human must still watch the road and be ready to take over.. For AI agentsAI agent, n.: An AI system that does not just answer but acts: given a goal, it decides on steps, uses tools such as a browser or code editor, checks the results and keeps going until the job is done. several schemes compete: some key the levels to the human's role, such as K. J. Kevin Feng, David McDonald and Amy Zhang's five levels (user as operator, collaborator, consultant, approver or observer; 2025); some to how much control flow the model decides, from fixed code through routers and state machines to fully autonomous agents; and Google DeepMind's levels of AGIlevels of AGI, n.: A scheme from Google DeepMind researchers for grading AI by how well it performs and how broad its skills are, from emerging to superhuman, much as self-driving cars are graded by levels. pair capability levels with separate autonomy levels. A recurring point is that autonomy is a deployment choice distinct from capability: a highly capable model can be run at low autonomy.
We ship the agent at the approver level: it drafts every change, but a human signs off before anything is merged.
Origin
Graded automation dates from Sheridan and Verplank's 1978 MIT report on human and computer control of undersea teleoperators. SAE J3016 (2014) made numbered 'levels' familiar for self-driving cars. Agent-specific schemes multiplied from 2023: Meredith Ringel Morris and colleagues' 'Levels of AGI' (Google DeepMind, 2023), LangChain's 'how agentic' spectrum (2024) and Feng, McDonald and Zhang's 'Levels of Autonomy for AI Agents' (2025).
Usage note
Levels describe a system's configured behavior, not its competence, and one agent can run at different levels for different actions — reading files freely but needing approval to send email. Numbering differs between schemes, so 'level 3' means different things in different frameworks.
Related
- levels of driving automationlevels of driving automation, n.: A six-step scale, from Level 0 to Level 5, describing how much of the driving a car's automation does and how much the human must still watch the road and be ready to take over.
- levels of AGIlevels of AGI, n.: A scheme from Google DeepMind researchers for grading AI by how well it performs and how broad its skills are, from emerging to superhuman, much as self-driving cars are graded by levels.
- human oversighthuman oversight, n.: Keeping people able to watch, understand, question and, if necessary, stop or overrule an AI system, rather than letting it act unchecked.
- human-in-the-loophuman-in-the-loop, adj.: Describes AI systems in which a person checks, corrects or approves the machine's work at key points, rather than letting it act entirely on its own.
- permission systempermission system, n.: The rules that decide what an AI agent may do on its own, what it must ask a person about first, and what it may never do — like a phone's settings for which apps can use the camera.
- agentic AIagentic AI, n.: AI built to pursue goals on its own — planning, taking actions and adjusting along the way — instead of only responding to one prompt at a time. 'Agentic' describes how much of this independence a system has.
- AI agentAI agent, n.: An AI system that does not just answer but acts: given a goal, it decides on steps, uses tools such as a browser or code editor, checks the results and keeps going until the job is done.
- capabilitycapability, n.: Something an AI system is able to do, such as writing working code or solving math problems, and how well. Measuring capabilities is how researchers track progress and spot dangerous skills.
Sources
- Feng, K. J. K., McDonald, D. W. & Zhang, A. X. (2025). Levels of Autonomy for AI Agents. arXiv:2506.12469. arxiv.org/abs/2506.12469v1
- Parasuraman, R., Sheridan, T. B. & Wickens, C. D. (2000). A Model for Types and Levels of Human Interaction with Automation. IEEE Transactions on Systems, Man, and Cybernetics, Part A, 30(3).
- Sheridan, T. B. & Verplank, W. L. (1978). Human and Computer Control of Undersea Teleoperators. MIT Man-Machine Systems Laboratory technical report.
Mentioned in
agentic AIagentic AI, n.: AI built to pursue goals on its own — planning, taking actions and adjusting along the way — instead of only responding to one prompt at a time. 'Agentic' describes how much of this independence a system has., AI agentAI agent, n.: An AI system that does not just answer but acts: given a goal, it decides on steps, uses tools such as a browser or code editor, checks the results and keeps going until the job is done., levels of driving automationlevels of driving automation, n.: A six-step scale, from Level 0 to Level 5, describing how much of the driving a car's automation does and how much the human must still watch the road and be ready to take over., autonomyautonomy, n.: How much an AI system can act on its own without a person approving each step, from merely making suggestions to carrying out long tasks with no supervision at all., human-in-the-loophuman-in-the-loop, adj.: Describes AI systems in which a person checks, corrects or approves the machine's work at key points, rather than letting it act entirely on its own., permission systempermission system, n.: The rules that decide what an AI agent may do on its own, what it must ask a person about first, and what it may never do — like a phone's settings for which apps can use the camera., levels of AGIlevels of AGI, n.: A scheme from Google DeepMind researchers for grading AI by how well it performs and how broad its skills are, from emerging to superhuman, much as self-driving cars are graded by levels.