The scale of AI right now is not on the model's side — it is on the human's side
Someone hears “AI is amazing” as “chat feels smooth”. Someone hears it as “that whole stretch of the process can be swapped out”. Someone hears it as “the supply structure and cost curve have to be re-laid”. All three are true, and they sit several layers apart. This site does one thing: measure where you stand using three layers of knowledge, then hand you a panorama you can read downward, layer by layer.
Three layers, and the order cannot be reversed
What you have done
The open questions ask what you actually did, not what you think. “I used it to draft my weekly report” and “I wired it into an approval flow” are two different worlds.
Why it can
A level is not a score. It is “the layer of mechanism you should understand”. Each level states four things: what it feels like, what you should know, where people usually go wrong, and the single next move.
Which part of the world it holds
The 63 nodes split into five zones: river, bridge, street, city, wild. Your level decides where you start reading — not which single zone you are allowed to see.
Why “level”, not “score”
Scores are for showing others; levels are for using yourself. A score makes people want to farm it higher. A level answers one question: which axis should you move next. So the result page gives no percentile ranking — only three things: which cell you are in, what the cell you are missing looks like, and where on the panorama that cell starts.
This is also the limit of the test: it measures “your cognitive coordinates on AI”, not “the ceiling of your ability”. Someone can sit at a low level and still do beautiful work — that only means their criteria live in their head, unsaid. Conversely, someone who can recite a pile of nouns but has never run one thing end to end is usually not high either.
Six levels
Listen
Heard the nouns, never touched it
Wade
Can ask and try, can't say why
Handle
Can wire and use, can land one small thing
Make
Can build your own tools and flows
Orchestrate
Can organise human–machine division, set criteria and boundaries
Set the board
Can change rules and supply, can anticipate the next layer
Start
About 5 minutes. For open questions write concrete things — words like “improve efficiency” are scored as empty. You can leave midway; nothing is saved. Coming back starts a new run.