Customer complaints: can AI express empathy that the team cannot deliver on?
A generative assistant can write a warm response to an unhappy customer even when the employee has limited means to resolve the problem. The management question is specific: how can managers prevent this expressed empathy from creating expectations of support that the team cannot meet?
What the literature establishes
Alicia Grandey (2003, Academy of Management Journal) distinguishes surface acting, which involves changing the emotion displayed, from deep acting, which involves working on the emotion actually felt. Her study associates these strategies with different patterns of emotional exhaustion and service quality rated by colleagues. It does not concern AI, but establishes a useful distinction: displaying the expected emotion does not fully describe either the employee’s experience or the service they provide.
Perceived authenticity is more than tone
Thorsten Hennig-Thurau, Markus Groth, Michael Paul and Dwayne D. Gremler (2006, Journal of Marketing) experimentally examine the effects of emotional displays in a service encounter. Their findings show that perceived authenticity contributes to the customer’s response beyond the mere presence of a smile. For an AI-assisted reply, this suggests a hypothesis to test, not an established finding: a caring tone does not guarantee a relationship perceived as sincere.
The premature conclusion
One might conclude that every expression of empathy should be written entirely by a human. These studies do not establish whether human-written or AI-assisted messages are preferable, and help with wording may still be useful. The managerial task is instead to examine the commitments implied by the text: do personal follow-up, priority handling or a quick resolution reflect actual capacity? (our executive and employee training programmes)
What the evidence does not establish
These studies predate generative AI and do not demonstrate that an assisted response damages customer relationships. Nor do they justify treating an in-person encounter and a written complaint in Italian as equivalent. A review limited to the message’s politeness would therefore miss the issue to be checked: any gap between the support announced and the support actually provided.
A practical check in Lugano
As part of SHR’s « Manager à l’ère de l’intelligence artificielle » programme, one exercise is to review, over two weeks, AI-assisted replies to complaints in Italian at a bank, fashion business, trading company or family-owned SME in Lugano. Before sending each reply, the team records every explicit commitment concerning follow-up, deadlines or action, assigning an owner and specifying the evidence expected to confirm delivery. At the agreed review date, it calculates the proportion of commitments fulfilled within the promised timeframe among those already due, retaining only the data needed for the check. This rate does not measure empathy; it checks whether the organisation delivers what its replies promise. To go further: explore the Managing in the Age of Artificial Intelligence training in Lugano in the canton of Ticino, or browse our executive and employee training programmes in Switzerland.
In pictures: Managing in the Age of Artificial Intelligence in Lugano
- managing in the age of artificial intelligence
- Lugano
- research
- training lugano