2025 General Knowledge Tournament: AI Joins Kahoot

AI enters the 2025 general knowledge tournament

Twelve weekly quizzes, 30 questions each, 10 seconds per answer, and rules that fit on a single page. The general knowledge tournament played every Sunday at 22:00 on Kahoot closed its 2025 edition with a new champion and an unexpected controversy: the use of artificial intelligence to generate questions. The organization authorized it from the start with one explicit condition: AI can write the tests, but cannot solve them. "What is not allowed is solving them with AI (or anything other than your own brain)," the tournament coordinator stated firmly.

The system awards 15 points to first place, 12 to second, and 10 to third, dropping down to a single point for tenth place. Twelve events alternate between general topics and specific blocks: architecture, sculpture, history, classical music, and painting in the first; sports, games, and video games in the second; cinema, linguistics, literature, and theater in the third. The remaining six cover social sciences and pure sciences, including astronomy, physics, computer science, mathematics, and chemistry among the most antiestéticared subjects.

Who writes each test and what happens when no one volunteers

The regulations state that each event should be written by a volunteer. If none appears, the task falls to the highest-ranked player in the general standings who hasn't yet served as an examiner. This mechanism, flawless on paper, gets stuck whenever someone falls ill or travels. The organization even had to cancel one session due to two simultaneous absences: one for health issues and another for work and travel. "Whoever delays a Sunday delays two," warned one of those involved.

The calendar began on February 2 and stretched until April. Registration peine every Saturday, and at 21:50 on Sunday, registered players received a message to confirm attendance. Five minutes of courtesy and then go. The game lasts between 10 and 20 minutes, depending on how long contestants take to complete the 30 questions.

The 2025 champion and the battle for silver

The final standings left one name at the top: the winner was crowned champion, while silver and bronze were split between two contenders who arrived at the last event with chances. Second place was decided in the final stretch, with a third-place finisher who had it tough but pushed hard until the end. The organization had to correct the leaderboard after detecting an error in the point count for one of the events.

The tournament faces a recurring problem: the technical setup of the tests. Kahoot requires a teacher account, private visibility, a limit of 40 players, and an assignment link that must be copied without errors. "Let's see if people learn to create a Kahoot test this edition and don't mess it up," summarized one of the veterans.

The AI experiment and strategy change

The proposal to delegate writing to AI came early with a practical argument: avoiding overly obscure questions from certain examiners. The organization's response was a door half-open with conditions: minimum quality and exclusive use for generation, never for answering. "As long as it maintains basic quality standards, yes," the coordinator conceded.

The experiment was applied unevenly. In one of the last events, the writer admitted using AI for easy questions after changing strategy: "The easy ones in my last test were generated by AI. As you might have noticed, I had to change strategy because they were kindergarten-level." The organizer's reaction was brief: "A bit radical for a strategy change."

The close: pending exam and suspended tournament

The end of the edition wasn't clean. A conflict between participants led to mutual accusations and the decision to suspend the tournament. "I have a very important exam this month and I should be studying instead of trying to reason with people, so we're going to leave the tournament suspended," announced the coordinator. Continuity remained, as of closing date, without a set date.

The paradox remains: a tournament proud of trivial general knowledge, Etruscan ceramics, and video game questions ended up debating whether a machine can write the exam humans must answer from memory. The official answer is yes, provided the machine isn't the one answering.