The Human’s Role in AI’s Fallibility

It's been a busy month for music and sports fans. July is gigging season, with musicians hitting the festival circuit and there's a host of blockbuster concerts to choose from. Arguably, the pick of the bunch – or at least for those fortunate enough to get a ticket – being the long awaited Oasis comeback.

July has also been a sporting bonanza, with the culmination of FIFA’s Club World Cup, the UEFA Women’s Euros and The British & Irish Lions tour in full swing and majors in golf and tennis with the Open Championship & Wimbledon.

Technology has played a key role in officiating all of these events – and will do so in NFL from later this year - but (so far) it has undoubtedly sparked the most controversy at SW19, where issues with the introduction of AI-powered Electronic Line Calling (ELC) have resurfaced the man v machine debate.  

The All England Lawn Tennis Club (AELTC) is no stranger to technology, having embraced Hawk-Eye’s ball tracking to improve line calling as early as 2007. In comparison, VAR wasn’t introduced in football until 2016. However, this year they took the leap, joining the Australian and US Open in going fully automated with their line calling. A lot was made in the media of Wimbledon turning its back on 147 years of tradition and there was scepticism both amongst the press and the players of the new system. AELTC would have hoped for some positive, high-profile interventions but sadly it went the other way and the sceptics were quick to point the finger at technology and cry controversy when a couple of errors were made.

There are a few things to dissect here. Firstly, mistakes by AI draw attention because expectations around accuracy are so high – in the same way that Carlos Alcaraz missing one easy forehand pass stands out more than the countless missed by players that do not possess his accuracy.

Secondly, the errors, while coming from the AI powered system, were of human origin. In one case, the operator accidentally turned off the cameras by mistake and in the other, the positioning of a ball boy as the player started to serve caused the machine to miss the start of the point and mistakenly call a shot as out thinking it was a serve.

Thirdly, in terms of officiating, AI is the logical extension of technology’s successful use to improve accuracy and impartiality in sport. Hawk-Eye systems offer a 99% accuracy. In comparison, historic research shows human line judges get approx. 8% of line calls wrong which equates to about four points per set. Hawk-Eye was first introduced to give players the chance to use technology’s greater accuracy to challenge these calls, in the hope of overturning the ones within the 8% they felt were wrong. While players efforts to correct errors are rarely successful – in 2024, less than 25% of the 1500+ line call challenges at Wimbledon were over-turned – ELC has proven very effective at doing so; the adoption of ELC at the US Open has reduced overturned line calls by more than 75% compared to human officiating.

The AI line judging at Wimbledon is essentially a sup-ed up version of the Hawk-Eye system previously used for challenges. We may miss the spectacle of player challenges and the drama of waiting for the ball tracking - where the wait for the path of the yellow arch is like a modern version of the gladiators wait for the emperor’s thumb – but the technology ultimately brings us less delays and more correct decisions.

Technology has also helped sport overcome the unintentional impartiality that comes with crowds. In a football study where 40 qualified referees reviewed incidents from a Liverpool home match against Leicester – half with crowd noise, half in silence – those with crowd noise awarded 15% fewer fouls against the home team than those watching in silence.

AI / technology was introduced to improve the officiating in sport – to remove the possibility of influence and reduce the margin of error. The key here is to reduce. Not remove. While AI cannot be swayed by a passionate, vociferous crowd baying for a decision, it can be affected by human interference or lack thereof as was the case at Wimbledon with aforementioned errant ball boys and click-happy operators.

Just as we humans have turned to AI to improve officiating, AI needs human involvement if it is to perform at its best and deliver the best sporting experience for both players and fans. Speed is a vital component of this experience – players and fans want fair and accurate decisions but not at the expense of the sporting spectacle. For example, at the FIFA Club World Cup the AI-powered semi-automated offside technology accelerated the decision-making process, reducing unnecessary play and giving match officials one less thing to worry about. At the tournament we also saw how VAR decisions, made quicker and with full transparency for the fans, were better received – something that must surely serve as a template for the technology’s use in domestic leagues around the world.

AI is fallible. It needs human safeguards. Mankind and technology need to work in tandem - to learn from their errors and improve the frameworks for success.

Or, as the reunited Gallaghers wisely put it, in their song Acquiesce – “…we need each other, we believe in one another”. Sport has shown it needs AI to improve officiating and AI needs people to function at its best. We now just need to believe in AI’s potential, rather than being blinded by its threats, and understand that ultimately, in the drive to improve sport, people and technology are in it together.  

It's been a very busy summer so far for the team at TEC, delivering the digital experience at leading sports events in golf and athletics, including the European Athletics Team Championships and U23 Championships. To find out more about the digital work we do for the governing body, check out our European Athletics case study