A proposed labeling system to categorize the level of AI use in blog posts and documentation, ranging from 0 (no AI) to 5 (AI-generated from prompts).

The scale includes five levels: AI-0 (human-written from scratch), AI-1 (AI used for translation), AI-2H (AI used for research but not written content), AI-2R (AI used to refine the author's own writing), and AI-5 (AI-generated from prompts). AI-3 and AI-4 are reserved.

The author characterizes AI-2 as reasonable for modern tech writing, while AI-5 represents "pure slop — Articles crafted almost entirely from a basic prompt."

The system is proposed as shorthand for online discussion about AI use in articles and as a voluntary disclosure mechanism for authors publishing content.