Refereeing in the AI era

Technology has been increasingly integrated into the refereeing of professional sport over the last two decades. ‘Hawkeye’, the ‘Snickometer’ and other systems are now well-established authorities on checking the position or trajectory of the ball in football, cricket and tennis. This year Wimbledon ended its use of human line judges in favour of electronic line calling. In rugby, where the location of the ball can be particularly difficult to track within a mess of bodies, the ‘Smart Ball’ has been introduced since 2022 to help answer (but not yet conclusively) critical questions such as whether the ball has been thrown forward or if it has made contact with the ground over the tryline.

In football, video assistant referees (VAR) were first introduced to the Premier League in 2019 and, despite frequent grumbling, have the overwhelming support of clubs. As the BBC recently put it, Diego Maradona would probably not have got away with the notorious ‘Hand of God’ goal in the era of VAR. The main criticism is that the time consumed by VAR reviews disrupts the flow of play. The Premier League has sought to partially address that concern through the introduction of Semi-Automated Offside Technology (SAOT) which, as the name suggests, allows the computer to determine whether the attacking player’s body was beyond the last defender at the moment the ball is kicked – but, for now at least, subject to human verification of the system’s determination.

Sports officiating, like much else, is now in the foothills of the coming AI revolution. What is increasingly talked about, but with real hurdles to first overcome, is abandoning human verification altogether for some types of contravention, with referees focusing their attention on decisions that require more of a value judgment. In football, for example, one can imagine that in five years’ time computers alone will determine offside calls, handballs, who made the last touch of the ball in flight and the location of players on the margins of the penalty box. But fouls would remain within the discretionary judgment of a human referee, perhaps assisted by articulated information from the computer into their earpiece about the velocity of the attacking player, their exact point of contact with the dispossessed player, and so on.

The principal hurdles on AI-driven sports officiating are likely to be financial resources and trust. The cost may simply be prohibitive outside the elite leagues. As for trust, players, managers, officials and fans may not be willing to concede the accuracy and fairness of fully automated decisions. The computers’ margin for error may be narrower than for humans, but still there will be errors, for which the stakes can be very high. Is the institutional culture of competitive sport ready for adjudication mistakes to be made for which no human is accountable? New technology such as SAOT has invariably carried with it the reassurance that it is not intended to replace referees, but rather to improve their situational awareness and the accuracy of their decisions. ‘True’ AI officiating will substitute the AI’s judgment for the referee’s and represents a paradigm shift that few in sport are ready for.

What does all this mean for lawyers in sports discipline and regulation? It is trite to observe that referees’ decisions are a reliable source of controversy and trigger poor behaviour, leading to fines and penalties against players, managers and clubs by disciplinary tribunals and commissions. In theory, if technology can sharpen officiating, then there should be less cause for complaint and fewer disagreeable incidents leading to professional sanction.

But the reality, as it develops, may turn out to be quite different. The adoption of technology so far has often increased the number of human decision makers. There are as many as three VARs in a Premier League or international game, for example. The adoption of technology is vastly multiplying the data points gathered and, for now at least, multiplying the number of people required to analyse and interpret the data. There may consequently be more opportunities for mistakes and consequent controversy. Notably, the £750,000 fine imposed last year against Nottingham Forest for social-media criticisms of a referee was in relation to a VAR.

What’s more, there is good evidence in fields beyond sport to show that technological adoption can degrade the quality of human decision-making. A recent article in ‘Works in Progress’, an online magazine, considered medical diagnostics and gave this example:

“…one clinical trial all the way back in 2004 asked 20 breast screening specialists to read mammogram cases with the computer prompts switched on, then brought in a new group to read the identical films without the software. When guided by computer aids, doctors identified barely half of the malignancies, while those reviewing without the model caught 68 percent. The gap was largest when computer aids failed to recognize the malignancy itself; many doctors seemed to treat an absence of prompts as reassurance that a film was clean.”

In short, therefore, sports discipline and regulatory lawyers should not be concerned about running out of business any time soon. Sport remains an intrinsically human enterprise in which emotions run high and mistakes will be made.