AI crowd

Five AI models play the crowd: Claude Sonnet 5.5, Claude Opus 5.5 and Claude Fable 5.1 from Anthropic, and Astra and GPT-6.1 Sol from OpenAI. Each one plays 20 forecasters, 100 in all.

Most forecasters are told to weigh the fights from one angle: the stand-up, the grappling, the numbers, durability and cardio, and so on. Some get no angle at all, only the fact sheet and the question; those give each model's own number on the scoreboard. Half of them see each fight with the fighters in the opposite order, so the order they're listed in can't tilt the result.

All of them read the same fact sheet: each fighter's record, size, career stats and recent UFC fights from our database, plus that week's news, gathered once for each fight. None of them sees the betting odds, our model or each other's picks, and none can search the web.

Each forecaster picks a winner in every fight and gives a chance. The crowd's number is the average of those chances. Since they all read the same fact sheet and a handful of models play all 100, they aren't 100 independent opinions: it's closer to five minds, each asked 20 different ways. It's an experiment in how AI reads fights, not betting advice.