When AI Assists at the Frontline: External and Internal Marketing Effects in the Context of Banking Mortgages

Artificial intelligence (AI) increasingly assists frontline employees (FLEs) in nonroutine tasks; however, its implications remain uncertain. This study examines how thinking AI, i.e., a machine learning-based tool that derives recommendations from historical patterns, affects external outcomes (market offerings and satisfaction with FLEs) and internal outcomes (FLE job autonomy and satisfaction with the responses received). Quasi-experimental evidence in banking reveals that AI-assisted FLEs adopt more conservative behaviors, resulting in decreases in mortgage approvals, default payments, and customer satisfaction with FLEs. Textual analysis and a survey provide first evidence consistent with AI-generated solutions appearing more rigid and less open to alternative possibilities, in turn reducing FLE satisfaction. A preregistered experiment tests this evidence, showing that AI-generated responses are perceived as less flexible, which in turn reduces their satisfaction with the response received and, primarily through this indirect pathway, constrains their perceived job autonomy, regardless of perceived message helpfulness. This study advances the marketing literature by showing that the role of AI at the frontline may not be inherently supportive for nonroutine tasks. AI should be carefully aligned with the characteristics of such tasks so that marketers can harness the potential of AI assistance while mitigating unintended effects on employees and customers alike.