Athletes’ Experiences of AI-Generated Images in Qualitative Research

Qualitative visual methods include approaches such as photo elicitation, photovoice, autophotography, and drawing, which can be used to build rapport, generate discussion within interviews, or as sources of data for analysis in qualitative studies. AI-generated images present a new frontier in visual qualitative methods: using images generated with AI programs can provide an innovative approach to creating images that represent participants’ experiences, but these have not been widely used or critically examined. The purpose of this study was to examine athletes’ perceptions of creating AI-generated images about their experiences of success and adversity in sport. The study included 10 participants between 20-61 years of age (8 women, 2 men) from various high performance sports (basketball, hockey, judo, rowing, running, swimming, track and field, ultramarathon). Athletes participated in qualitative interviews about their experiences of success and adversity in sport. During the interviews they worked with the researcher to generate images about their experiences using the AI program Midjourney, and athletes reflected on the images that were generated. Thematic analysis of the interviews generated the following themes: (a) Inaccuracies and mis-representation; (b) Misembodiment and disconnection with images; (c) Over-representation of emotional experiences; (d) Generating diverse representations of athletes; and (d) Re-creating important moments. Athletes also discussed challenges and limitations, as well as potential positive uses for AI-generated images in qualitative research. This study demonstrates how AI-generated images are different from other visual qualitative methods, and the results provide insights for exploring athletes’ experiences using AI-generated images in qualitative studies.