AI Identifies Tennis Players’ Emotions with 68.9% Accuracy, Raises Ethical Concerns: KIT Study

Researchers at the Karlsruhe Institute of Technology and the University of Duisburg-Essen have developed an AI model that can identify emotional states from the body language of tennis players with up to 68.9% accuracy. The AI was trained using real-life footage of tennis matches, learning to associate body language signals with different emotional reactions. The study, led by Professor Darko Jekauc, found that both humans and AI are better at recognizing negative emotions. The technology could have applications in sports training, healthcare, education, and customer service, but raises ethical concerns around privacy and data misuse.

AI in Sports: Recognizing Emotions in Tennis Players

Researchers at the Karlsruhe Institute of Technology (KIT) and the University of Duisburg-Essen have developed an artificial intelligence (AI) model capable of identifying affective states in tennis players based on their body language during games. This study, published in the journal Knowledge-Based Systems, marks the first time AI has been trained with data from actual games to assess body language and emotions. The AI model demonstrated an accuracy comparable to human assessments, but the study also raised ethical considerations.

The interdisciplinary team of sports scientists, software developers, and computer scientists developed a unique AI model. They utilized pattern-recognition programs to analyze video footage of tennis players during actual games. The AI model was trained using real-life scenes, a significant departure from the simulated or contrived situations typically used in AI training. The researchers focused on the body language displayed by 15 tennis players when they won or lost a point. The AI model was then trained to associate these body language signals with different affective reactions.

AI’s Success Rate in Identifying Emotions

The AI model developed by the researchers demonstrated a success rate of 68.9 percent in identifying affective states. According to Professor Darko Jekauc of KIT’s Institute of Sports and Sports Science, this accuracy is comparable, and sometimes superior, to assessments made by human observers and earlier automated methods. The use of real-life scenes for AI training is a significant advancement in identifying real emotional states and making predictions in real scenarios.

Interestingly, the study found that both humans and AI are better at recognizing negative emotions. Jekauc suggests that negative emotions might be easier to identify because they are expressed more obviously. Psychological theories propose that humans are evolutionarily better adapted to perceive negative emotional expressions, as quickly defusing conflict situations is essential for social cohesion.

Potential Applications and Ethical Considerations

The study envisions numerous sports applications for reliable emotion recognition. These include improving training methods, team dynamics, and performance, and preventing burnout. Other fields such as healthcare, education, customer service, and automotive safety could also benefit from the early detection of emotional states.

However, the potential risks associated with this technology, particularly those relating to privacy and misuse of data, must be considered. Jekauc emphasized that their study strictly adhered to existing ethical guidelines and data protection regulations. He also stressed the importance of clarifying ethical and legal issues before implementing such technology in practice.

Conclusion

The study by KIT and the University of Duisburg-Essen represents a significant step forward in the use of AI in sports science. The AI model’s ability to accurately identify affective states from the body language of tennis players during games could have far-reaching implications for training methods and performance enhancement. However, the ethical and legal implications of such technology must be carefully considered before its widespread adoption.