Inauthentic inclusion: Exploring how intention to use AI‐generated diverse models can backfire

Abstract Rapid advances in AI technology have important implications for, and effects on, brands and advertisers. Increasingly, brands are creating digital models to showcase clothing and accessories in a similar way to human models, with AI used to customize various body types, ages, sizes, and skin tones. However, little is known about how the underrepresented consumers respond to a brand's intention to use AI‐generated models to represent them. We explore this by conducting four studies. We find evidence that a brand's intention to use AI‐generated (vs. human) models negatively affects brand attitude (study 1). We further investigate this effect using two different underrepresented consumer groups: LGBTQIA+ consumers (study 2) and consumers with disabilities (study 3). We show the effect to be serially mediated by consumers' perception of greater threat to their self‐identity and a lower sense of belonging, subsequently having a negative effect on brand attitude. Finally, we show that the perception of a brand's motivation for representing diverse consumer groups can attenuate these negative effects (study 4). Specifically, when consumers believe a brand is intrinsically motivated to use AI‐generated diversity representations, they report a significantly lower social identity threat which in turn is associated with a significantly higher sense of belonging to the brand. Our research findings suggest that a brand's well‐meaning intentions to represent diversity can in fact have negative effects on the very consumers whom a brand is trying to attract. While catering to diversity is of critical importance, our results indicate that brand managers should exercise caution when using AI to appeal to diverse groups of potential consumers.