Trust or Autonomy? Delegating Purchase Decisions to Autonomous AI Agents

Abstract

Objective: This study investigates consumers’ willingness to delegate purchase decisions to autonomous AI agents in e-commerce. The model examines a trust-based facilitating mechanism and a restraining mechanism arising from Perceived Loss of Autonomy. Perceived AI Quality, Perceived Usefulness, Social Trust, Cognitive Trust, Affective Trust, and Perceived Loss of Autonomy are examined in relation to Purchase Decision Delegation Intention.

Method: An online cross-sectional survey collected data from 328 experienced adult online shoppers. Seven constructs were measured using 23 items. Age, Gender, Online Shopping Frequency, and AI-Assisted Shopping Frequency were included as controls. PLS-SEM in SmartPLS 3 was used to assess the measurement model and structural relationships.

Findings: Reliability, convergent validity, and discriminant validity were supported, and all hypotheses were confirmed. Perceived AI Quality was positively associated with Cognitive Trust (β = 0.424, p < 0.001), which was positively associated with Affective Trust (β = 0.401, p < 0.001). Affective Trust had the largest direct coefficient for Purchase Decision Delegation Intention (β = 0.301, p < 0.001), followed by Social Trust (β = 0.249, p < 0.001) and Cognitive Trust (β = 0.180, p = 0.001). Perceived Loss of Autonomy was negatively associated with PDDI (β = -0.150, p = 0.001). The model explained 41.6% of PDDI variance.

Conclusion: Consumers’ readiness to delegate purchase decisions reflects a balance between trust in AI agents and preservation of perceived autonomy. Effective autonomous shopping agents should foster Cognitive and Affective Trust while preserving meaningful consumer control and choice.