I’m trying to be clearer about how I use AI on this site. Different posts get different kinds of help, so I have started adding an AIL label to them.
The AI Influence Level (AIL) framework, created by Daniel Miessler, uses a scale from 0 to 5:
- AIL 0: Written without AI
- AIL 1: AI helped with grammar or phrasing
- AIL 2: AI made significant edits or expansions
- AIL 3: AI wrote from a human outline
- AIL 4: AI wrote from a basic human idea
- AIL 5: AI generated nearly everything
What This Looks Like Here (Right Now)
Most posts on this site are AIL 0 or AIL 1. I write them myself and sometimes ask Claude to check the grammar or point out an unclear explanation.
This post is AIL 2. I made the outline, Claude drafted parts of it, and I edited the result. AI did meaningful work on the prose.
For technical posts, I sometimes use AI to add explanatory comments to code snippets. I count that as AIL 1.
I’ll update this section as my habits change. The lastModified date shows when I last revised it.
Why I’m Trying This
If AI did meaningful work on a post, I think that is useful context for the reader. This scale is an experiment, and I may find that it is too broad or too specific.
I use Claude Code for development, regular Claude for editing and explanations, and GitHub Copilot for completion while I type.
There is a difference between asking AI to check my grammar and giving it an outline to turn into prose. The badge is a short way to describe that difference.
How This Might Change
AI tools are useful for specific tasks and oversold for plenty of others. At Propel, I use them to parse messy government websites and identify error patterns.
My usage will change, and the framework itself may stop feeling useful. I’ll update this post when that happens.
An AIL badge tells you what kind of help was involved in that post.