Résumé
The last decade has seen a rapid increase in the use of artificial intelligence (AI) approaches to statistical analysis. The most common of these, machine learning (ML) and deep learning (DL), have become commonly used in the sport sciences. In this presentation, we will discuss a recent rapid review we conducted to summarize and critique current sport science research using these technologies in athlete development contexts (i.e., athlete development, talent identification, and athlete selection). The review indicated that ML and DL approaches were prominent in three areas: improving athlete assessment, athlete selection and classification, and athlete development and training. However, the research questions under investigation and analytical approaches used generally reflect a continuation of established ways of thinking in sport science, rather than new, paradigm-shifting developments. There is considerable potential in the use of ML and DL, since these approaches allow researchers to access more data, more easily, and with greater statistical complexity, than ever before. However, the indiscriminate use of these technologies has the potential to perpetuate well known biases in research designs and analyses. Ultimately, a balanced approach that embraces both innovation and critical evaluation is necessary to ensure these tools enhance, rather than disrupt, the athlete identification, selection, and development landscape.
