The feeling skills gap: the role of empathy in voice-driven AI for service recovery

As companies increasingly deploy voice-driven AI to enhance service efficiency, its impact on perceived customer orientation remains underexplored. Drawing on AI Constraints and Feeling Economy perspectives, this research examines when and how voice AI affects perceptions of customer orientation in service recovery. Results show that customers perceive providers as less customer-oriented when recovery is handled by AI (vs. humans), due to AI’s tendency to reduce emotions into quantifiable parameters (parametric reductionism) and fail to convey empathy. Mediation analyses confirm that perceived empathy explains the link between agent type and service outcomes. We identify a boundary condition: the negative effect of voice AI emerges only when there is a task–ability mismatch, in which the task requires feeling skills that the AI cannot convincingly deliver, but not when the task calls for thinking skills, for which AI’s capabilities are aligned with task demands. We further discuss implications for AI in service recovery.