Human-only
Made for human brains.
AI can now write essays, solve math and pass exams in seconds. That makes one question more important than ever: what can you do on your own?
Whole Board exists to answer it. It's built to keep AI from taking the test for you. Every score on our leaderboards is meant to be one person, thinking alone, against the clock.
How we built it
Making a test that's both fair and hard to cheat took a lot of work. Before writing a single question, we:
- Researched across the web: published intelligence research, the CHC model that modern IQ tests are built on, how clinical tests like the WAIS are structured, and open research projects such as the International Cognitive Ability Resource (ICAR).
- Studied what makes online IQ tests feel fake: flattering instant scores, recycled viral questions, paywalls at the end, and one-number results. Then we designed around every one of them.
- Wrote original questions and question generators. We didn't copy any test's items. Pattern, Number, Spatial, Logic, Memory, Speed and Focus questions are built from rules every time you start, and Word Power draws from our own reviewed bank.
- Tested the generators automatically, at scale: 400,000 random tests across 50 separate runs, with independent solvers re-solving 1.6 million grids, number chains, shapes and logic puzzles to confirm each has exactly one correct option and belongs in its category. Word Power questions were reviewed one by one for a single clear answer.
The result is a test that's simple if you put your thought into it, honest about its accuracy, and built for the way this generation actually thinks.
Why this matters now
AI assistants went mainstream in just the last few years, and habits are shifting with them: how much we memorize, how long we focus, how often we work a problem out ourselves instead of asking a machine. Whole Board is built to measure how people think in that world.
Whole Board is building a new human baseline. As people take the tests, we recalibrate what "average" means using real players' first attempts. Over time, your score tells you where you stand against people living in the same AI-filled world you are, not against people tested in a different era.
That only works if the scores are real. One person pasting questions into a chatbot doesn't just cheat the leaderboard. It corrupts the baseline for everyone.
Every rule has a shield
Rules that only ask nicely don't stop anyone. So every rule here is backed by something the site does on its own, on our server, where nobody can switch it off.
| The rule | What enforces it, automatically |
|---|---|
| Your brain only. No AI, search, calculators, notes or help. | Server-side clocks sized for one focused human; questions drawn as images, not text; fresh questions every attempt; tab-leave and time-away limits; inhumanly fast answers get unranked. |
| One person, one ranked account. | Ranking needs a Stripe Identity check (government ID + live selfie). One identity can only ever rank one account. |
| No bots or scripts. | Automated browsers are detected and unranked. Every answer is re-scored on our server from the original seed; the browser never sees an answer key for anything that counts. |
| No restarts to study questions. | The timer starts on our server once and can't be restarted. Closing or reloading ends the attempt. |
| No double-dipping. | Two tabs, two clicks or two devices at the same moment can't create a second score, a second refund or an overfull team: the database locks the record and only the first one counts. |
| Rank where you live. | Verified players can move their home board once per season, and a place only shows on the boards once 5 verified players live there. |
| Teens with teens, adults with adults. | Ranked boards and Squads are 18+. Private matches only let teens join teens and adults join adults. Teen profiles are only visible to other teens. |
| Money never buys rank. | No purchase touches a score. Prices are set on our server, not in your browser, and refunds follow fixed rules that run automatically. |
| Nobody hand-edits a score. | Scores come only from the scoring code. There is no "edit score" button anywhere, including for the founders. |
| Gaia can't help you cheat. | The site guide refuses test questions, disappears on every test and game page, and goes silent everywhere while you have a test running. |
How we keep it human
- Tight, server-side timers. The clock starts on our server the moment you press Begin. Each question gets enough time for a focused person to think it through. Stopping to ask an AI eats into that time, and leaving the tab is detected.
- Questions are drawn, not written. Questions appear as images instead of page text, so they can't be copy-pasted or read as page text by AI browser tools.
- Fresh questions every attempt. Pattern, Number, Spatial and Logic puzzles are generated when you start, so there's no answer key to look up. Word Power draws a new mix from a rotating bank.
- Tab-switch detection. Leaving the test once gets a warning, in case a notification pulled you away. Leaving more than 1 time, or being away for more than 4 seconds total, removes the attempt from the leaderboards.
- Automation detection. Browsers that report being automated, and inhumanly fast answers, are flagged and unranked.
- Real-time tasks. Memory, Speed and Focus move in fractions of a second, leaving little room for outside help.
- No restarts. Closing or reloading mid-test ends the attempt, so questions can't be studied and retried. You're warned before you start and again if you try to leave. Accidents happen, so we automatically restore one interrupted attempt every 30 days, including returning the credit for a paid one.
What happens if an attempt breaks the rules
The attempt is saved in your history with the reason, but it doesn't count toward any leaderboard or your overall IQ. Repeated or deliberate cheating can get an account removed from the leaderboards or banned. See the Terms.
Accessibility
Questions are drawn as images so bots can't read them. That would shut out blind and low-vision players, so Word Power, Number Sense, Logic and Memory have an audio mode: questions and options are read aloud, you answer with the number keys, and you get extra listening time calculated from each question's length. Pattern, Spatial, Speed and Focus are visual by nature, but the four audio categories are enough to earn an overall IQ.
Honest limits
No online test can watch your room. Someone determined could use a second device. Our job is to make that slow, risky and not worth it, and to keep the leaderboards clean. If you see a score that looks impossible, report it and we'll review the attempt's timing data.