Research on diversity, equity, and inclusion (DEI) research in the workplace uses human images to explore biases related to gender, sexual orientation, and race. However, natural human images often lack standardization and fail to adequately represent diverse identities, limiting the reliability and generalizability of findings. Variability in image quality and contextual factors, such as lighting and background, further complicates the interpretation of results. This study addresses these methodological challenges by using Stable Diffusion, a deep learning model, to create AI-generated images as standardized stimuli for DEI research. These images were standardized for technical consistency—ensuring uniform lighting, resolution, and facial expressions—and evaluated for realism, likability, trustworthiness, and competence through a pilot survey with 500 participants. The final dataset comprised 20 images deemed inclusive and realistic. Results revealed no significant differences in participants’ evaluations of warmth, competence, and professionalism across AI generated photos, contributing the reliability of AI-generated images.
