🤖 🫱🏻🫲🏾 🧠 What is QANTA 2025?
Welcome to QANTA: the world’s first Human–AI Cooperative Trivia competition!
We’re building a fun, interactive battlefield to discover who really reigns supreme in question answering—humans, AI systems, or a perfect team-up.
The only thing missing is you!
How Can You Participate?
🤖🛠️ Build an AI Teammate: Design an AI system that makes accurate guesses with clear explanations, and just the right amount of confidence to help your human partner shine.
🧑🤝🧑 Play as a Human: Bring your knowledge, strategy, and intuition—then team up with an AI agent to outthink the competition.
✍️📝 Write Challenging Questions: Author the quizzes that will put teams to the test (and get paid for every question we use!).
No fee to join. QANTA is all about science, community, and curiosity!
💡 New this year: Team up with AI for "bonus" questions and more collaborative gameplay!
🤔 Why Join?
AI is really good at answering questions—except when it isn't—can they know they don't know?
Our competition digs deep to find out:
For Tossups:
📊 Test Calibration: Can models judge their own confidence accurately?
🤖 Compare AI & Human Judgment: Who’s better at knowing what they know?
🏅 Climb the Leaderboard: See how your system stacks up using our special calibration metrics.
For bonuses:
🔺 Boost Human Performance: Do AI explanations actually help people score higher?
📝 Question Submission Deadline: June 7, 2025
📚 Half-Packet Deadline: June 1, 2025 (for writers submitting a full set)
🤖 System Submission Deadline: June 7, 2025
🧑💻 Player Registration Deadline: June 10, 2025
🏟️ In-Person Tournament (College Park, MD): June 14, 2025
🌐 Online Tournament: June 21, 2025
✍️ Want to Write for QANTA?
👉 We’re seeking creative question writers!
👉 Submit at least 10 challenging tossups and bonuses on any theme.
👉 Earn $5 for every question accepted!
💪🏼 The questions will be hard, at least 3-dot and each packet will be themed so as to be about topics that computers struggle on.
🖥️ We provide an interface for writing the questions. We highly encourage using it; it uses AI-in-the-loop to help construct questions computer stumbles on.
🧩 The target length is around 450 characters.
We currently are in discussions with authors to create theme packets on:
Military history
World literature
Spatial reasoning
And more!
If you have any questions, please contact us at qanta@googlegroups.com.