Human–AI teams for decision-making in banking and finance: A systematic scoping review

Purpose: This systematic scoping review synthesizes empirical evidence on human–AI team decision-making in banking and financial services, a domain characterized by rapid AI adoption but limited documentation of how that adoption shapes collaborative decision processes. Methods: Following the Arksey and O’Malley framework and PRISMA-ScR guidelines, three databases (Scopus, Web of Science, PsycINFO) were searched, along with snowball sampling. Thirteen empirical studies met eligibility criteria. Results: Three collaborative architectures were identified: human-as-final-decision-maker, human-as-supervisor, and iterative co-construction with AI consistently operating as the initial analytic agent. Trust dynamics ranged from calibrated reliance to algorithm aversion and commitment bias. Explainable AI supported sensemaking, but only when it was designed interactively. Organizational incentive alignment and AI literacy emerged as decisive moderators. Conclusions: In the included studies, performance gains from human–AI teaming in finance were evident but depended on collaboration architecture, trust calibration, and sociotechnical design. Human-centered outcomes and task performance were partially independent dimensions of effectiveness. Empirical documentation of AI deployment in banking remains thin and geographically concentrated, warranting longitudinal field research and empirically grounded governance standards.