Precisely comments on AI reliability in sport

​​​​​​​​​​​​​

This year’s Football World Cup is set to see greater use of AI-powered technologies to enhance match analysis capabilities, drive fan engagement with officiating technologies such as using semi-automated offside technology (SAOT) – which will include producing 3D avatars of all players – as well as in-game and player analysis to support countries competing.

But with the debate about the use of SOAT continuing after its efficiency was brought into question during a football match last year, we ask Dave Shuman, Chief Data Officer at Precisely what organisers can do to ensure the accuracy of AI-driven decisions ahead of major tournaments?

“When referees depend on real-time AI guidance in high stakes matches and tournaments, the data pipeline effectively becomes part of the officiating team.  That means the information feeding these systems must be accurate, consistent, complete, and contextual every single time.  SAOT has previously shown both its potential and its vulnerabilities.  For example, the match delay in 2025 – where officials couldn’t rely on the system for several minutes – highlights that the weakest link is often the data pipeline, not the model itself.  Reliability is not just about speed; it’s about delivering accurate, consistent and explainable decisions under pressure.”

“The most significant risks to the use of AI in the upcoming World Cup are rooted in data quality and operational complexity.  For example, biased or incomplete training data can lead to inconsistent or unfair outcomes across different leagues, play styles or athlete profiles.  Similarly, variability in stadium environments including lighting, camera placement and sensor infrastructure can affect data capture quality.  On a global stage, these risks are magnified, and the need for AI-ready data becomes greater.  A single inaccurate call can influence results, media narratives and public trust.”
 
“Ahead of the tournament, organisers should strengthen data integrity by validating every data source at the point of capture – whether camera feeds, sensors, or tracking systems – before it reaches an AI model.  They should also implement continuous, real- time data quality monitoring to detect latency, sensor drift, missing frames, or anomalies immediately.  Maintaining strong governance and lineage as well as contextual consistency will also ensure that systems understand match dynamics, player movement and environmental factors.  Finally, transparency is crucial when implementing AI to guarantee that decisions can be clearly explained and to maintain trust amongst players, coaches and fans.”