AI service robots in physical retail: Can they reduce the intrusiveness of human retail assistants and prevent shoppers from leaving the store early?

Although in-store shopping has traditionally been viewed as enjoyable, recent industry reports suggest a contrasting reality: consumers increasingly describe it as socially taxing and psychologically stressful. Our preliminary study empirically supports this observation, demonstrating that consumers frequently experience stress—particularly perceived intrusiveness—in physical retail environments during interactions with human retail assistants (HRAs). Importantly, our findings further indicate that this sense of intrusiveness is largely driven by perceived judgment, which in turn triggers emotion-focused coping responses (prematurely leaving the store). Building on these preliminary findings and drawing on Transactional Stress Theory, we conducted five experimental studies in fashion and skincare retail contexts to examine whether artificial intelligence retail assistants (AIRAs) can disrupt this stress process and serve as a more comfortable alternative to HRAs: retail assistant type (HRA vs. AIRA) → perceived judgment → perceived intrusiveness → emotion-focused coping (premature store departure). We also investigated whether service design cues (the presence vs. absence of smile) and consumers’ shopping goal clarity (low vs. high) function as boundary conditions within the proposed serial mechanism. Across five studies, the findings consistently support the proposed serial mediation effect (H2), as well as a direct effect of retail assistant type on store departure (H1). Moreover, the results reveal significant moderated mediation effects involving retail assistants’ smiling behavior (H3) and consumers’ shopping goal clarity (H4). Taken together, these findings suggest that AIRAs can alleviate an often-overlooked source of stress in physical retail environments—namely, the social-evaluative judgment and perceived intrusiveness embedded in human service interactions. These findings offer important implications for both theory and practice.