Using AI-generated Images to Minimize Researcher Bias and Enhance Participant Insight in Qualitative Interviews

Visual elicitation methods are used in qualitative research interviews to prompt discussion among participants and enhance researcher understanding. These methods involve the use of photographs or other visual stimuli to guide research interviews. With advancements in artificial intelligence, AI-generated images offer a promising way to enhance consistency and ensure ethical compliance with photo elicitation techniques. The purpose of this study was to explore the use of AI-generated images as a visual elicitation method in qualitative interview guides. In our study, researchers used the online platform DALLE-E to create 36 AI-generated images depicting men and women across six ethnicities with characteristics previously found to be intimidating in exercise environments; ‘fit’, ‘muscular’, and ‘lean’ individuals (2 gender x 6 ethnicities x 3 characteristics). Image prompts were created using various filters and control words to ensure consistency. During qualitive interviews, participants were asked a series of questions regarding the factors that contributed to them being intimidated by others in fitness facilities and the specific characteristics of others that they perceived as most intimidating. Near the end of the interview, participants were asked to pick out which images appeared visually intimidating and explain why they perceived the images as intimidating. The use of AI-generated visuals provided participants with an opportunity to clarify their perspective and distinguish between characteristics they did and did not perceive as intimidating, thereby reducing researcher bias. Additionally, the images facilitated deeper participant responses and insights by allowing comparisons between images and expansion of their initial verbal responses.