Human-AI Autonomy Scale

About the scale

From AI in the void to Human in the void.

AI in the void

AI is absent or irrelevant to the workflow.

AI in the loop

AI assists, but does not act independently.

Human in the loop

AI executes, human supervises and approves.

Human in the void

AI operates independently, humans are not required.

AI in the void · Human baseline

00 / Human-led

You do the development work without generative AI.

AI does

None. Compilers, linters, CI and ordinary IDE tools still count as human-led work.

You decide

Choose, write, check and ship the changes.

For example: You implement a fix, run tests and review the diff yourself.

2022-2026 / observed workflows

How this sample changed

Mean inferred minimum level of linked, self-selected posts. Not the average developer or a population trend.

2026 / 25 accounts

4.96 / 12

Mean observed floor, not maximum capability.

Typical practice in this sample

Parallel agents and background bug fixes coexist with advice-only use and manual review.

Models actually named

Opus 4.6, GPT-5.4, DeepSeek v4 Pro, Astra and Luna named; most unspecified.

Independent coding evaluation

SWE-bench Verified leaderboard (2026): Feb 2026, mini-SWE-agent v2.0: Opus 4.5 resolved 76.8%; Gemini 3 Flash and MiniMax M2.5 each 75.8%.

External comparisons, not results for these accounts. Tasks and setups differ across years; live leaderboards may change. Scores are not autonomy levels.

25 dated source posts for 2026
1 / level 12 / level 83 / level 54 / level 25 / level 56 / level 57 / level 58 / level 69 / level 510 / level 711 / level 712 / level 513 / level 514 / level 215 / level 516 / level 517 / level 518 / level 519 / level 520 / level 521 / level 522 / level 523 / level 324 / level 825 / level 5

Reading the scale

A level describes one workflow, not a person or a model.

About the Scale

Most frameworks ask: “What can the AI do?” This one asks: “Who is doing the work — the human or the AI?”

The Human-AI Autonomy Scale is a developer-first framework that measures ownership of execution and decision-making in software development workflows.

Other frameworks measure model capability; this one measures how work is actually performed in practice.

Same tools. Completely different reality.
Two teams can use the same assistants and still operate at very different autonomy levels.

  • Hiring clarity: “This role operates at level 6–7.”
  • Team alignment: “We want to move from level 4 to 6.”
  • Expectation setting: avoid “we use AI” mismatch.
  • Strategic planning: map transition paths between levels.