Maintaining the Integrity of Sport – Time for AI to Take the Lead ?

The February 6th decision of the Gymnastics Ethics Foundation Disciplinary Commission in relation to allegations of score manipulation against Ms. Evangelia Trikomiti and European Gymnastics at the 2024 Rhythmic Gymnastics European Championships (RGEC) raises further questions over the role of humans judging/supervising subjective sporting outcomes and whether it is time for AI to play a more prominent role.

The Facts

In May 2024, the RGEC took place in Budapest. It was the last possibility for one European individual to qualify for the Paris 2024 Summer Olympic Games. The main contenders for Olympic qualification were Ms. Liliana Lewinska (POL), Ms. Panagiota Lytra (GRE) and Ms. Vera Tugolukova (CYP). Ultimately, Ms. Tugolukova qualified for the Games, securing 0,25 points more than Ms. Lewinska.

Gymnastics has a range of procedural safeguards designed to preserve the legitimacy of its judging processes. The Superior Jury is an important part of the gymnastics integrity framework. It essentially supervises competitions, the decisions/scores of judges panels, and the overall results system. Ms. Trikomiti served as President of the Superior Jury at the RGEC in Budapest.

Following the RGEC, Ms. Trikomiti was charged with acting partially in favour of the Cypriot gymnast Ms. Tugolukova against her rivals for Olympic qualification, especially Ms. Lewinska. It was claimed that,  in breach of various GEF ethics, and the Judges Oath, Ms.Trikomiti had manipulated the competition by unduly interfering with the judges’ work, to ensure that the athlete from Cyprus, obtained the Olympic place.

A Contested Hearing  

In evidence before the Commission, the legitimacy of Ms. Trikomiti’s conduct was questioned in several areas. It was pointed out that the number of blocked scores for Ms. Lewinska, in particular in the ball exercise, where all judges’ panels were blocked, was exceptional. It was also alleged that Ms. Trikomiti did not explain the reasons for the blocks but intervened in the scores in favour of Ms. Tugolukova or against Ms. Lewinska. 

In her defence, Ms. Trikomiti contended that she did not interfere with or manipulate the scoring of the RGEC and had not infringed any of the applicable rules in relation to judging, scoring, blocking or governance . She further noted that , although she was not obliged to do so, she did not judge Ms. Lewinska, Ms. Tugolukova and Ms. Lytra herself, leaving it to the TC vice-president. 

In short, the defence was based on the contention that Ms. Tugolukova achieved the Olympic place, rather than Ms. Lewinska, because, based on legitimate interpretations of the Code of Points applied by qualified and experienced Superior Jury members, Ms. Tugolukova was entitled to receive and did receive higher scoring for her performances.

The Human Dimension

The scoring for Rhythmic Gymnastics has necessarily subjective elements. It is not possible to use objective measurements, such as distance or timing. Despite the tight scoring protocol adopted in gymnastics, and reliance on a sophisticated results system, judges can have different views. Such views will inevitably be influenced by their personal interpretations of certain routines, music appreciation and sporting socialisation. The margins between qualification and elimination can be very small so tiny deviations can have a significant impact. 

Matters are further complicated by the close network of relationships that exist in elite sport, escalating nationality tensions and perceived resentment between individuals generated by previous experiences or assumed future ambitions.  This was illustrated before the Panel when it was alleged that one of the witnesses harboured resentments towards Ms. Trikomiti stemming from the 2012 London Olympic Games, as she purportedly believed that Ms. Trikomiti was responsible for her failure to qualify for those Games. 

Although the Panel found this explanation rather far-fetched, not least given the time that had elapsed, the line of argument demonstrates the challenge of maintaining trust /confidence and eliminating perceptions of self-interest in judging. It was also insinuated during the case that another witness might have ambitions to become the new President of the Technical Committee and was motivated accordingly.

Who wins ?

Ultimately, based on the evidence heard/received by the Panel, including the testimony of ten witnesses, the Panel concluded it was more likely than not that Ms. Trikomiti influenced the Superior Jury to alter scores in favour of Ms. Tugolukova (or to the detriment of Ms. Lewinska and Ms. Lytra), thereby enabling Ms. Tugolukova to secure the Olympic place. 

The decision is not final and can still be appealed to the GEF Appeal Tribunal, with a possible further appeal to CAS. At this stage, however, it is very difficult to identify any winners in this scenario. It appears that the gymnast who should have participated in the Paris Games was unfairly denied the opportunity. Who knows if she will have another chance and how her career may be impacted by the experience. The gymnast who did compete in Paris ought not to have done and is now likely to be remembered for all the wrong reasons.  

Witnesses who gave evidence to expose the misconduct and uphold the integrity of the judging process have been accused of bias, score settling and furthering their own ambitions within the sport. A long-standing, senior sports administrator has been found to have manipulated the score of a major qualification event and deemed to have left an impression of untrustworthiness. The case has taken over six months to conclude and been subject to various procedural delays which have increased costs. Confidence in the judging and scoring system in gymnastics has been undermined in and outside the sport.

The Future – A Better Way?

Is it time to look at an alternative approach to such matters? One possibility is to explore the merits of a system in which AI takes the lead in judging rather than operating as a mere supporting or back up function. Relevant sports could look to create and test potential AI solutions to specifically address the challenges generated by subjective assessment and decision making.  

This could conceivably involve a first instance decision being taken by an AI tool with the potential for human challenge to the AI determined outcome, albeit in narrowly defined circumstances. These challenges could be heard on a highly accelerated basis enabling any potential sporting injustices to be rectified in real time and well before major events take place. Such a framework would see an end to the type of personal challenges around the motives and credibility of witnesses that arose in this case.

Yes, this is easier said than done and the relevant AI tools will need to be created (and seen to be created) with caution to avoid the type of biases that have plagued human judging in the past. Provided, however, that there is sufficient transparency in the development of the AI tool(s) and the establishment of thresholds for challenges (which should be a rare occurrence) then an AI driven approach could potentially deliver more genuine impartiality, swifter, cost-effective sporting justice, and better maintain the true spirit of sport than the current human driven processes on offer.