Sports Journalists Asked Microsoft’s Copilot to Predict World Cup Matches, and the Results May Surprise You
TL;DR
USA Today sports writers asked Microsoft Copilot to predict four 2026 World Cup matches: Spain-Cabo Verde, Belgium-Egypt, Uruguay-Saudi Arabia, and Iran-New Zealand. Copilot predicted four wins, including a 3-0 Spain victory and narrow 2-1 or 1-0 results in the other games. All four matches actually ended in draws: Spain-Cabo Verde 0-0, Belgium-Egypt 1-1, Uruguay-Saudi Arabia 1-1, and Iran-New Zealand 2-2.
Nauti's Take
Copilot did not secretly understand football here; it likely retold the obvious: favorites win, underdogs resist for a while, and the final score looks tidy. That kind of neatly phrased average prediction is risky because it sounds like analysis.
For newsrooms, this is useful as a reality check, but weak evidence that AI can reliably forecast real sports events.
Briefingshow
The issue is not that Copilot missed a few score predictions. What matters is that the explanations sounded plausible while likely echoing favorite-based logic and media hype instead of modeling real match uncertainty. That is the risk with AI forecasting: weak assumptions can arrive dressed as confident analysis.
Sources