Sports Journalists Asked Microsoft’s Copilot to Predict World Cup Matches, and the Results May Surprise You
TL;DR
USA Today asked Microsoft Copilot to call four World Cup matches: Spain-Cape Verde 3-0, Belgium-Egypt 2-1, Uruguay-Saudi Arabia 2-1, and Iran-New Zealand 1-0. All four matches finished as draws. Copilot did not just miss the scorelines; it appears to have treated a draw as barely possible, despite soccer making that outcome routine. The sharpest miss was Spain versus Cape Verde: Copilot expected Spain’s attack to overwhelm a weaker defense, but goalkeeper Josimar Vozinha Dias helped force a 0-0 result.
Nauti's Take
This is less a shocking AI failure than a useful reality check. Copilot seems to have predicted the storyline, not the match: favorite strong, underdog weak, scoreline plausible.
That is where everyday AI becomes risky, because it sounds like an analyst rather than a guesser. These tools can help with research, context, and scenario framing, but forecasts should be labeled as speculative entertainment.
Briefingshow
The episode exposes the gap between persuasive analysis and actual forecasting. LLMs can narrate sports convincingly, but they often turn media narratives, team reputation, and training-data patterns into overconfident probabilities. For journalism, betting-adjacent products, and fan tools, the warning is simple: fluent reasoning is not calibrated uncertainty.
Sources