Eight scenarios. Swipe toward your answer.
Paper due in six hours. AI states a confident-sounding fact. A friend asks where to eat. Pick the move that's actually you, not the one that sounds smarter.
The three-minute test
Forty archetypes. Forty-one swipes. Three minutes. You get your AI IQ, your archetype, your blind spots — and what to do about them.
You don't just use AI. You relate to it. Some people command it. Some argue with it. Some defer to it. Some hide it. The pattern is sitting in every conversation you've ever had with the thing — you've just never seen it laid out. Three minutes, and you will.
How it works
Every question is a swipe. Forty-one cards in three kinds: pick one, yep or nope, and three where one answer is actually right. Together they put you on nine dimensions, and the nine dimensions decide the archetype.
Paper due in six hours. AI states a confident-sounding fact. A friend asks where to eat. Pick the move that's actually you, not the one that sounds smarter.
Right for yep, left for nope. Asked AI what to text back. Made a song with it. Argued with it because it was wrong. This is where the badges come from.
Two prompts, same goal. Swipe toward the one that works. It's the only part of the test you can get wrong, and it's how we separate people who can make AI perform from people who think they can.
I check its facts. I've used it to sound like I know a topic I don't. There's work of mine AI isn't allowed near. Nobody sees your answers but you, and the honest one is the useful one.
AI is more like a tool, or a friend? The you that uses it is the same you everyone sees, or a version nobody else does? Then four optional taps of context, and your archetype.
What it measures
Every answer moves you along at least one of these. Your archetype is where all nine land together. Drag the first four and watch the archetype change.
Who starts the work — you, or the model?
Do you check it, or do you trust it?
Are you here to understand it, or to be done with it?
One domain, or everything?
Can you actually make it perform?
A tool you pick up, or part of how you think?
Does it get the polished you, or the real one?
Are you building something with it?
Is your AI life private, or something you talk about?
The library
Twenty-four core archetypes describe how you relate to AI. Twelve specialty archetypes describe what you use it for; they show up as badges, or take over when they're the strongest thing about you. Four outliers catch the extremes. Open any of them.
What you get
The number is the headline. The rest is the point.
“I'm not building with AI. I'm building AI into something.”
For classes, teams, and communities
One link. Everyone takes the test. You get the room's archetype map: where the Architects are, where the Ghosts are hiding, and what to talk about the next time you're all together.
Under the hood
Every answer is logged with its response time against a random ID that lives in your browser. No name, no email, no login. The mini test is calibrated to ±6 points. The full test — a hundred questions — narrows that to ±2 and adds peer percentiles by major, year, and tool.
The test is the instrument for a student research project: two college sophomores studying how their own generation uses AI, and how it doesn't.
About the projectA random ID in your browser is the only thing that connects your answers. Nothing to log into, nothing to leak.
Built for the first generation that grew up with this. It works for everyone else too; it's just not written for them.
Responses inform the research and the next version of the test. They never feed a model.
The three-minute test is calibrated to ±6. The full test is a hundred questions: ±2, percentiles against your major, year, and tool, and a profile you can download. Leave an email and we'll tell you when it opens.
You're on the list. We'll email you when the full test launches.
That doesn't look like an email address.