Skip to content
outthinkAILog in
A round, end to end

How to play

  1. 01Pick a subject and a difficulty.
  2. 02You get one question — a real one, asked by a real person on Stack Exchange.
  3. 03Write your answer. Fifty words, hard cap.
  4. 04AI gets the same question and the same fifty words.
  5. 05A judge scores both against a fixed rubric, without being told whose is whose.
Questions people ask
  • Do I need an account?

    Yes. Ratings, streaks and leaderboards are tied to it. Playing is free, and you cannot buy extra rounds.

  • How long does a round take?

    As long as you want — there’s no timer. The verdict isn’t instant, though: judging runs in the background and the page waits with you.

  • Who is the judge?

    AI, given a fixed rubric for your subject, shown both answers blind in a random order. Four scored dimensions, five in Medicine, always adding to a hundred. The model you played can’t referee its own round.

  • What if the judge gets it wrong?

    It sometimes will. A verdict that fails the code checks is re-run, and if it fails again the round is set aside for a human and your day is handed back. Treat a verdict as a well-argued opinion, not a grade.

  • Can I play more than once a day?

    No. You cannot buy or earn a second round. It resets at midnight your time.

  • What happens to a question I don’t win?

    It can come back later, and you won’t be told it’s a repeat. Questions you win are retired.

  • When do I show up on a leaderboard?

    After five rated rounds. Rounds that break or get set aside don’t count towards it. Leaderboards are only visible to players who are signed in.

  • Does difficulty make it easier?

    It changes which question you get, and nothing about your opponent or your rating.

  • Where do the questions come from?

    The Stack Exchange network, under a Creative Commons BY-SA licence. We show the title only, and every question links back to the original post, its author, and the licence that applies to it.