When AI Recommends, Who Decides? Exploring User Control and Decision-Making in AI-Generated Financial Recommendations

This study examines how AI-generated financial recommendations and digital interface design influence users’ trust, perceived control, decision confidence, and willingness to follow financial guidance. The study begins from a simple but important question: when an AI system recommends what a person should do with their money, who still feels responsible for the final decision? AI can process large amounts of information quickly and can present users with recommendations that appear clear and useful. However, the ease of following a recommendation does not necessarily mean that the recommendation has been understood, evaluated, or consciously accepted by the user. This makes the interface through which an AI recommendation is communicated an important part of the decision-making experience. The proposed study uses a user-centered experimental approach with prototype digital interfaces that present the same type of AI-generated financial recommendation in different ways. Participants will complete simulated financial decision-making tasks and will be asked to evaluate the recommendation and make a final choice. Their experience will be examined through measures related to trust, perceived control, decision confidence, clarity, usability, and willingness to act on the recommendation. The study is designed to compare how different presentation approaches affect the movement from receiving AI guidance to making an independent decision. The study does not assume that the purpose of an AI interface should be to increase acceptance of its recommendation. Instead, it considers appropriate decision support as an interaction in which the user can understand the recommendation, question it when necessary, compare it with alternatives, and retain a meaningful sense of control. Existing research has separately examined trust in AI, over-reliance, recommendation acceptance, autonomy, explainability, and financial AI adoption. This research brings these concerns together from an interface and user-experience perspective. The expected contribution is a clearer understanding of how interface choices can support AI as a helpful guide without making the user feel that the decision has already been made for them.