In a bid to enhance fairness in sports technology, Wimbledon has introduced significant updates to its artificial intelligence systems. The All England Club, renowned for hosting the prestigious tennis tournament, has implemented these changes to minimize bias in AI-generated video highlights.
Traditionally, the club has relied on AI to analyze extensive match footage from its 18 courts, automatically selecting standout moments. The AI’s selection criteria have included player gestures, such as fist pumps, and audience reactions, like cheering. These curated highlights are then made available online for fans and players, offering a comprehensive review of performances.
This year’s tournament, concluding on July 19, has seen the AI system refined to address any potential bias favoring players with more expressive gestures or larger fan bases. The objective is to ensure that equally talented but less demonstrative players receive comparable recognition. “A great shot doesn’t always come with a fist pump,” noted Sam Seddon, a client executive at IBM, the club’s technology partner.
The issue of tech bias has gained prominence as awareness grows that AI systems, far from being impartial, often mirror societal biases due to their human programming. More severe instances include facial recognition errors with darker skin tones and financial algorithms imposing higher interest rates on Latino and Black borrowers. In response, some U.S. lawmakers advocate for stricter oversight, proposing an “algorithmic accountability” test for high-risk AI systems used by major corporations.
IBM’s Watson AI, which powers the video analysis, evaluates footage based on crowd reactions, player gestures, and the significance of each point, assigning an “excitement ranking” to identify top clips. These scores are now adjusted to mitigate bias, ensuring that all noteworthy moments are captured. “Without these adjustments, some of the best moments might be overlooked,” Seddon explained.