AI Use Scale
This scale describes how much AI contributed to a piece. It is meant as process disclosure, not as a judgment of quality or seriousness.
| Level | Description |
|---|---|
| 0 | Written before LLMs were invented. |
| 1 | Written after LLMs existed, but with no AI assistance. |
| 2 | AI used only for spelling, grammar, formatting, or transcription cleanup. |
| 3 | AI used for light copy edits, title ideas, or small wording suggestions. |
| 4 | AI used for brainstorming, outlining, or research direction, while the prose remained human-written. |
| 5 | AI used for limited draft text, examples, or local rewrites; the human still wrote or substantially rewrote most of the piece. |
| 6 | AI used for substantial drafting of some sections; the human determined the argument, structure, and final wording. |
| 7 | AI drafted a majority of the first version from detailed human notes or prompts; the human revised heavily. |
| 8 | AI drafted nearly the whole piece; the human edited, checked, and approved it. |
| 9 | AI produced the piece with light human steering, selection, and cleanup. |
| 10 | Written by AI with no human intervention and minimal prompting. |
The intent is to make the workflow legible at a glance. A short header label such as 2/10 or 7/10 is enough for most posts, and ai-notes can add extra detail when needed.