How Al Jazeera used AI to predict two out of three Fifa World Cup matches

The Idea

The 2022 Fifa Football World Cup was held in the Arab world for the first time in the tournament’s history. To mark the occasion, Al Jazeera wanted to give its audience a taste of the sights, sounds and emotions of the tournament across its digital platforms.

The Brief

Al Jazeera created an artificial intelligence robot named Kashef that analyzed wins, goals and Fifa rankings from the tournament’s history to predict the outcome of games. The AI tech used over 200 metrics combing through over 100,000 records to see who would win the following day. The records were run through Google Cloud’s Vertex AI, which produces state-of-the-art models, adopting Google Brain and using its Efficient Neural Architecture Search (ENAS) to find the best prediction model. The data pipeline was built with GCP-BigQuery.

After every match, the model was rerun to predict the outcome of the next game all the way through to the final. Predicting match results is tricky when external factors like team morale or player fitness make a big difference in how the game goes. While most other publications ran predictions games over the tournament, Al Jazeera’s use of AI made it more accurate with a success rate of 67%. Readers were encouraged to ’take on the robot’ to see if their predictions were better than Kashef.

Along with the prediction game, Al Jazeera’s coverage included a daily updated match center, video commentary of each day’s matches, daily quizzes and social media campaigns, infographics, data visualizations and galleries of the history of the World Cup.

The coverage was built around user interaction and gamification delivered on multiple platforms in English and Arabic.

The Results

Al Jazeera managed to rack up more than half a million plays of the predictor game.

This project won at The Drum Awards for Online Media 2023. Find out more about The Drum’s Global Awards program or head to our case studies hub to read more award-winning stories.