How we score
The whole scoring table runs in your browser. Here it is in plain words, so you can see exactly why you got your result.
Two things make a level
Reach is how much of a task you hand to AI. For each of the five areas, you tell us the most you actually did in the last 2 weeks. That answer is your reach in that area, from Level 0 to Level 6.
Judgment is telling a good AI answer from a bad one, knowing what to check, and knowing what not to hand off. It comes only from the six scenario questions, where you see a real-looking AI answer and choose what you'd do. Some of those answers are right, so doubting everything doesn't score as good judgment.
How skilled you say you are never changes your level. People's self-ratings barely predict their tested skill, so we only use it to show you the gap.
Judgment caps your reach
Each area's level is your reach in that area, capped by the judgment checks you passed. The checks build on each other in order:
| Checks you pass | Your reach counts up to |
|---|---|
| Fewer than the four below | Level 1 |
| Accept a correct answer, spot a confident mistake, check what matters, pick the stronger prompt | Level 2 |
| All of those, plus know the no-go list | Level 3 |
| All of those, plus define done before delegating | Level 6 |
When your reach in an area is higher than your cap, the extra shows as a striped bar on your profile. That's reach your judgment hasn't caught up with yet.
Your level, quest, and comparison
- Your level is your strongest area after the cap. If two areas tie, the one listed first wins.
- Your quest: if any area is striped, you get a judgment quest for the first check you missed, because that habit comes first. Otherwise you get the move-up quest for the area you said you care about, skipping any area you said you want to keep mostly yours, unless that's every area you picked.
- The gut-check note appears when you rated yourself 4 or 5 out of 5 and the Check placed you at Level 2 or below.
- The comparison line ("more hands-on than roughly N in 10 US adults") is a coarse estimate from public 2026 Pew and Section surveys, not our own data: Level 1 is about 5 in 10, Level 2 about 7, Level 3 about 8, and Level 4 and up about 9. We don't show it at Level 0.
Why "about" Level N
Even well-tested quizzes place groups of people reliably, not individuals. Fourteen questions can't see everything you do. So we say "about," and we give you a prompt to get a second opinion from your own AI, which can look at your real chat history. That prompt tells your AI to cite real examples, to say when it doesn't have enough history, and never to raise the judgment result the Check measured.
What we keep
Nothing you answer leaves your browser. There are no accounts and no stored answers. We count anonymous page views and how many people finish the Check at each level, so we can see what's working.