Book Review: AI in sports – the good, the bad, and the incomprehensible

By: Anne Tjønndal, Daniele Canini and Stian Røsten

AI in sports – the good, the bad, and the incomprehensible

The use of Artificial Intelligence (AI) is becoming increasingly integrated into various regions across the sporting landscape. AI’s capacity to analyse vast amounts of data, identify patterns and ultimately help to make predictions and decisions, is often anticipated to drive advancements in sports, movement and health. The application of AI in these three domains’ manifests, for example, from the identification and development of athletes, providing coaches game strategy recommendations and injury predictions, assisting referees in decision-making, to enhancing the fan engagement. The book Artificial Intelligence in Sports, Movement, and Health reflects the ongoing academic and public interest in the ways in which AI might transform sport, movement and health. Particularly, the aim of the book is “to empower readers with knowledge and enhance the understanding of the transformative potential of AI in sports, movement and health” (p. v). While the editors Carlo Dindorf, Eva Bartaguiz, Freya Grassmann and Michel Fröhlich note that it is important to bridge the “gap between theory, science and practical applications […] to realise the full potential of these technologies” (p. vii), we would argue that it is equally important to critically examine the (often unintended) implications of AI in sport, movement, and health. However, the book offers little to no engagement with such critical reflections on the use of AI.

______

One of the main issues concerns the absence of an introduction and a conclusion that would provide an overview and accompany the reader through a coherent reading of the individual chapters.

The book consists of five main parts, comprising a total of 15 chapters, which provide diverse contributions of the role and potential AI hold for the three realms stated in the title: “sports, movement and health”. After a brief editorial section outlining the aim and structure of the book, readers are left with the five main parts. Part I, Digital Transformations: Artificial Intelligence’s Role in Sports Science (chapter 1-3), delves into the broader realm of AI’s role in sports science. In chapter 1, Johannes Lenard begins by situating AI into the broader field of sports science and explore the potential impact this may have from a philosophical point of view. Next (chapter 2), Richard Latzel and Patrick Glauner explores how AI applications like ResearchRabbit (e.g., to review existing literature within a field), Elicit (e.g., to summarise research papers), DeepL (e.g., to translate text), to name some, can assist the academic research and writing processes.  In the final chapter (3) in this part, Tessa Menges explores the application of AI in endurance sports, encompassing aspects such as training optimisation, race selections and strategies.  Part II, Artificial Intelligence in Medical and Health Aspects of Sports (chapter 4-6), moves on to explore AI’s role and impact in the realm of medicine and health. It includes three chapters examining diverse topics on fall prevention for older adults (Wolfgang Kemmler), sports-related injuries predictions (Robin Owen, Julian A. Owen and Seren L. Evans) and anti-doping analysis (Maxx Richard Rahaman and Wolfgang Maass). Part III, Human-Computer-Interaction (chapter 7-8), addresses the human-centred design principles of AI aiming to enhance the user engagement in performance analysis, injury prevention, and personalised healthcare interventions (Marco Speicher and Patrick Berndt) and how to visually present sporting data by using, for example, heatmaps, error bars and probability distribution (Christina Gilmann).

In Part IV, Motion Capture and Feedback Systems (chapter 9-10), Bernd J. Setter and Thorsten Stein begin by exploring the applications of machine learning (i.e., supervised, unsupervised and reinforcement learning) for biomechanical analysis of human movements to prevent injury, treat diseases and enhance performance. This includes examples of how machine learning (ML) models can be trained to classify knee injury status based on patterns of athletes’ muscle activation. In a similar vein, Melanie Baldinger, Kevin Lippmann and Veit Senner, in the following chapter, explores how markerless motion capture technologies like Depht Camereas can be used to create a three-dimensional image of athletes’ bodies to analyse and optimise the performance of various athletic movements, such as running, jumping and rowing. They argue that these technologies can be used to identify abnormalities in movements relevant to enhance performance or assist rehabilitation processes. Finally, in Part V, Practical Examples of Machine Learning and Predictive Analytics (chapter 11-15), diverse examples showcasing AI’s potential to shape the future of sports is given. This includes examinations of AI’s impact on tennis performance analysis (chapter 11), predicting sporting outcomes (chapter 12), enhance recreational marathon running through individualised training programs and race strategies (chapter 13), performance optimisation in football (chapter 14), as well as in talent identification and development processes within youth sport contexts (chapter 15).

For readers interested in the development of AI technologies in sport, movement, and health, several chapters in the book offer insights into these topics. In particular, chapters 6, 10, 13 and 14 provide insight into the cutting-edge development of sports technologies, though the presentation may be somewhat technical. While some of the individual contributions in the volume offer an interesting read, the book as a whole exhibits certain weaknesses that warrant attention. One of the main issues concerns the absence of an introduction and a conclusion that would provide an overview and accompany the reader through a coherent reading of the individual chapters. Without a shared outline, framework or a final summary, the book must be read mainly as a collection of individual articles, rather than a cohesive volume on artificial intelligence in sports, movement and health. This absence makes the book feel disjointed at times. It is difficult to identify a coherent vision and coherence between the chapters. It is also challenging to identify a logical or thematic progression between the sections. Despite being divided into thematic sections, each chapter appears self-contained, with few cross-references or meaningful connections to other contributions in the book. A clear example of this is the repeated inclusion of concept definitions (particularly of AI, deep learning (DL), and machine learning (ML)) at the beginning of multiple chapters. This repetition creates a sense of redundancy that we find unnecessary

Another relevant issue concerns the stylistic and content-related disparity between chapters. Some contributions are written in a descriptive or reflective style with a narrative tone, while others adopt a highly technical approach, using specialised academic language. This lack of consistency makes the book difficult to follow and raises questions about its intended audience. It is unclear whether the book is aimed at an academic readership, sport professionals, or technology developers. Furthermore, the book’s overall approach to AI in sport, movement, and health comes across as strongly techno-optimistic. Many of chapters tend to emphasise the potential of artificial intelligence, with little recognition of its unintended implications or ethical dilemmas (such as privacy and regulations concerning sport performance and health data). Some notable exceptions are chapters 5 and 6.

As scholars with a background in the social part of sport science, we found it difficult to read and absorb the empirical contributions presented in the book. These seem to be aimed at readers with a background in technology, engineering, biomechanics and/or medicine. There are some interesting discussions of AI in sports, movement and health presented throughout the book. However, a greater editorial effort aimed at building a strong coherence between the individual chapters would have been appreciated by this trio of readers. Furthermore, an introduction clearly outlining the key concepts of the book (AI, DL, ML) could have worked to eliminate the conceptual redundancy found in several chapter introductions, avoiding repetition of concepts. We also believe that a more in-depth discussion of ethical, social and cultural issues would broaden the scope of the book, making it more relevant for scholars interested in the social implications of AI in sports, movement and health.