Should AI or human agents handle customer complaints? Exploring the impact of agent type and complaint response type on recovery outcomes

Firms increasingly use AI agents for consumer complaints due to their speed and cost-effectiveness, but consumers are often reluctant to accept them. Understanding when to use AI agents and which response strategy to adopt is crucial. Drawing on intentionality theory, we examine the interaction between agent type (AI vs. human) and response type (apology vs. denial) on recovery outcomes (i.e., forgiveness, repurchase intention, and attitude). Across four experiments, we find that human-delivered apologies and AI-delivered denials lead to more favorable outcomes. These effects are parallelly mediated by perceived sincerity and fairness. We further identify boundary conditions: anthropomorphism enhances recovery when AI agents apologize but backfires when they deny; human agents are more effective in addressing multiple complaints from the same consumer; and prior denial from an AI agent improves the effectiveness of a human agent’s denial. Our findings offer theoretical insights into AI-based complaint handling and provide practical guidance for firms.