Questions about AIDM.

Why not percentages?

Generated code is edited, reorganized, reviewed, or rewritten, while planning and debugging may materially involve AI without mapping to lines. Workflow and agency are more honest units.

Why not numeric levels?

The modes are ordered by AI agency, but AIDM attaches no numbers to that order. Numbers imply meaningful distances, invite averaging, and drift into being read as scores. Named modes keep the order useful without becoming a rating.

Why would a project declare honestly?

Because it helps the project. Contributors learn what workflow and review to expect, adopters can calibrate their own review, and speculation about AI use is replaced by a statement the project controls. No mode is ranked or colored as better, so there is nothing to gain by understating AI involvement.

How does this relate to commit trailers, PR disclosure, or SPDX?

Commit trailers and pull-request disclosure describe one change. SPDX's AI profile describes AI models shipped inside the software. AIDM describes the typical workflow of the whole project. They complement each other, and AIDM requires none of them.

Is Autonomous worse than Assistive?

No. Modes are not a ranking. They say how work was organized, not whether the software is good, safe, maintainable, or trustworthy.

Do I have to name the AI tool?

No. Tool and model disclosure is optional; the specification is intentionally vendor-neutral.

What if contributors work differently?

Use the dominant substantive workflow if one clearly exists. Otherwise use Mixed or publish scoped declarations.

Is this an AI detector or certification?

No. AIDM is a self-declared transparency convention and does not verify the declaration independently.

Who is behind the project?

AI Development Mode is an open project created as part of Astzweig's Social Contribution efforts and intended to evolve through community feedback.