AI is increasingly embedded in industrial service recovery processes. Yet its use in industrial contexts, especially regarding its potential dark side consequences for customer relationships, remains underexplored. This paper addresses this gap through a three-study investigation. Study 1, a field survey, shows that greater AI involvement increases customer over-reliance, which in turn reduces trust and customer engagement. Study 2, an experiment, demonstrates that customers penalise failures attributed to AI involvement more harshly than human-caused failures. Our analyses show that this penalty can be reduced when governance mechanisms are in place, particularly when human oversight is visible (e.g., human-in-the-loop interventions), with similar but smaller effects observed for algorithmic explanations. Study 3, a two-wave field survey over a 24-month period, shows that over-reliance evolves into complacency, which subsequently undermines trust and engagement, and that governance safeguards weaken this longitudinal pathway. Together, these findings advance theory by integrating automation bias, algorithm aversion, and relational governance into a unified process model of AI-enabled service recovery, showing that AI shapes customer relationships through both behavioural and attribution-based pathways over time. For managers, the results highlight that AI does not automatically enhance relationships; instead, firms must actively design governance mechanisms that maintain oversight, transparency, and accountability. • AI involvement in industrial recovery increases customer over-reliance. • Over-reliance evolves into complacency, eroding trust and engagement. • AI-attributed failures trigger stronger fairness penalties than human errors. • Human-in-the-loop and explanations as governance safeguards mitigate AI-related justice penalties. • Governance safeguards curb long-term relational erosion in B2B recovery.
Vik Naidoo (2026) studied this question.