Why this page is specific
This instrument asks you to be precise about your own rigour. It would be indefensible to be vague about ours, so this page says what was generated, from what, and who signed it off, rather than gesturing at "AI-assisted".
The level descriptors
The whole authority of this instrument rests on 100 level descriptors: 20 dimensions, five levels each, describing what a 1 looks like as opposed to a 2, concretely enough to be checked rather than agreed with. Here is how they are made.
- The source is not a model's opinion. Every descriptor is drafted from the long-form article for that dimension in the AI Readiness series. Those articles were written with AI support as well, for research, drafting and editing. What makes them Jurgen De Smet's is not the typing: the observations come from his client engagements, the argument and the judgement are his, and he approved every article before it was published. The model is never asked what it thinks readiness looks like. It is asked to compress material that already exists.
- AI drafts. A named human edits. Drafting is done with AI. Every descriptor is then read, corrected and approved by Jurgen De Smet before it reaches this site. None of them ship on the strength of being plausible.
- Nothing scales before it is reviewed. The instrument is built one cluster at a time, and the format is locked on real, human-reviewed content before the rest is written. That sequencing exists precisely so that a hundred descriptors cannot be generated in one pass and quietly shipped.
Where the words come from
Three steps, and AI is in two of them:
- 01The observation. Human, and not reproducible by a model. What organisations actually do, seen across client engagements over years.
- 02The articles. Written with AI support for research, drafting and editing. Directed, corrected and approved by a named human before publication.
- 03The descriptors. Drafted by AI from those articles, then read, corrected and approved one at a time before they reach this site.
If that sounds like a lot of AI, it is. Claiming otherwise would be the easy thing to write and the wrong thing to publish, on a page whose only job is to be accurate about exactly this.
The rest of the site
AI is used extensively across the writing and research behind this site: for analysis, for arguing with an idea until it breaks, for drafting and for editing. It is also used to write the code this site runs on.
The opinions, the arguments and the model itself are ours, not a model's. Where a factual claim is made, the underlying source is checked by a human rather than trusted to a language model. Nothing on this site is published under a byline that did not read it.
What is not AI-generated
Your score. The instrument does no inference and makes no judgement about you: it takes the levels you chose, reports the lowest one, and shows its arithmetic. There is no model between your answers and your result.