In the BBC's 2026 World Cup predictor game, users have demonstrated a greater ability to forecast match outcomes than both BBC Sport's football expert Chris Sutton and an AI prediction model. This trend was particularly evident in the second round of group games, where user predictions showed significant improvement.
Cape Verde, a team that has surprised many with its performances, saw 99.65% of users predict a loss against Spain and 83% predict a loss against Uruguay. Despite these underdog scenarios, the collective user performance outpaced expert and AI predictions. In the second round of 24 group games, users achieved 18 correct results, an improvement from 13 in the first set of fixtures. Chris Sutton's predictions yielded 14 correct results in the second round, an increase from his performance in the initial games. The AI, generated using Microsoft Copilot Chat, improved from 13 correct results to 15 out of 24 in the second round.
Chris Sutton is providing predictions for all 104 games in the tournament, as well as ranking the predicted order of finish for each of the 12 groups. The BBC predictor game allows participants to select a winner for each match or opt for a draw. The final set of group matches, including games like Scotland versus Brazil and England against Panama, will further test the predictive capabilities of users, experts, and AI.