AI-Generated Visual Design and User Behaviour on Social Media: Practitioner and Institutional Perspectives from Bahrain

Abstract

Generative artificial intelligence (AI) is now embedded in the production of social media visuals. However, research has concentrated on production processes rather than audience outcomes, and empirical work in the Arab context remains scarce. This study examines how AI-generated visual design is adopted by creative professionals, deployed by institutions, and understood by those practitioners to affect user behaviour. Nine semi-structured interviews were conducted in Bahrain with three purposively sampled groups: content creators, institutional representatives, and specialists. Audience response was therefore accessed through practitioner accounts rather than through direct user data. Interviews were analysed thematically in MAXQDA against a ten-theme framework, yielding 223 coded paragraphs. Three key findings emerge. Adoption is task-contingent rather than scalar, ranging from no reliance on purely graphic work to 80% reliance on institutional video, with time-saving the dominant motivation while claims about creativity are consistently hedged. Verification of AI output is not a single ethical competence but a family of domain-specific capacities derived from each institution's own accountability regime. Practitioners across sectors and languages independently reported that content depicting people outperforms AI-generated alternatives, suggesting that established attention and identification mechanisms constrain substitution. Claims concerning users are practitioner-reported and require corroboration through direct audience research.