How AI is changing gymnastics judging
AI-powered computer vision systems are being introduced into elite gymnastics judging to measure elements like body angles, rotations, and handstand positions more consistently than human judges. Commenters weigh potential benefits—reduced national and personal bias, clearer difficulty scoring, and use in training—against risks such as entrenched algorithmic bias, athletes “gaming” the model, loss of artistry, and the Goodhart-style distortion that occurs when complex human performance is reduced to a rigid point system. The conversation broadens to similar tensions in figure skating and team sports refereeing, questioning whether technology can or should replace inherently subjective human judgment in aesthetic sports.
Current Judging & Why AI Is Being Introduced
- Gymnastics already uses a detailed 200+ page Code of Points, with difficulty and execution panels plus reference judges; scores get three decimals from averaging multiple judges.
- Human judging suffers from bias and limited viewpoints (e.g., missing a 15° off-vertical handstand that should be penalized).
- AI/computer vision is pitched as a tool to measure body positions and angles more consistently, not (yet) to judge artistry or complex connections.
What “AI” Is in This Context
- Several commenters argue this is really “computer vision,” rebranded as “AI” for marketing.
- Others reference the “AI effect”: once something works reliably, people stop calling it AI.
Fairness, Bias, and Consistency
- Supporters: a single standard system, usable in training and competition, is more consistent than rotating human judges with varying preferences and national biases.
- Skeptics: models can encode new biases (e.g., body type, race, unusual sizes), be brittle to lighting/camera changes, and people may wrongly assume they’re objective.
- Some suggest AI should augment human panels, with humans able to override suspect calls.
Gaming the System & Goodhart’s Law
- Strong concern that athletes will optimize to “what the AI sees,” just as they currently optimize to human blind spots.
- Comparisons to:
- Wii Sports–style “minimum movement” to trigger sensors.
- Pitch framing in baseball.
- Figure skating’s IJS system, where complex jumps and base values + GOE led skaters to chase points, degrading perceived “artistry” and variety.
- General Goodhart theme: once a metric becomes the target, it stops measuring the intended “greatness.”
Artistry vs Technical Precision
- Debate over whether sports like gymnastics/figure skating should be closer to:
- Pure technical difficulty and biomechanical precision, or
- Artistic, hard-to-formalize “greatness.”
- Some welcome a more technical, less artistic direction; others fear “robotic,” soulless routines optimized for micro-deductions.
Analogies to Other Sports and Tech Refereeing
- Suggestions to use similar systems in fencing, HEMA, martial arts, and big leagues (NBA/NFL) to reduce referee errors.
- Counterpoint: over-precise, always-on enforcement (like some uses of VAR or strict speed limits) can harm game flow, fan experience, and miss the “spirit” of rules.