aceModerate evidence
More
Humanlike chatbots anger customers more when service goes wrong, experiments find
Study title: Mitigating Chatbot Service Failures: The Roles of Anthropomorphism, Acknowledgment, and Apology
- Evidence type
- Controlled or validated comparison
- Sample
- n = 1,124
- Sources
- 1 sourcePeer-reviewed
- Published
- 2 min read
- Truvace record
- TRV-2026-2651cite or check it
Study title: Mitigating Chatbot Service Failures: The Roles of Anthropomorphism, Acknowledgment, and Apology
Researchers ran three pretests (30, 120 and 240 participants) and two scenario-based experiments (346 and 778 consumers) in which people evaluated service interactions with chatbots designed to seem more or less humanlike. The experiments found that humanlike design made customers more aggressive when the failure concerned the interaction process itself.
Read the study abstract
ABSTRACT This study examines how chatbot anthropomorphism shapes consumer responses to service failure by distinguishing between process failure and outcome failure and by testing alternative recovery responses. Across three pretests ( N = 30, 120 and 240) and two scenario‐based experiments (Study 1: N = 346; Study 2: N = 778), consumers evaluated interactions with high‐ versus low‐anthropomorphic chatbots in service settings. The findings show that chatbot anthropomorphism intensifies customer aggression under…
59Gain side
How the two sides scoreClose call
56Problem side
The gain
In two scenario experiments (346 and 778 consumers), an apology after a process failure reduced customer aggression more than acknowledging the chatbot's AI nature, especially for highly humanlike chatbots.
The problem
In two scenario-based experiments (346 and 778 consumers), highly humanlike chatbots intensified customer aggression after a process failure because they violated stronger expectations of polite, competent interaction.
A fixed formula sets both scores from the evidence and its sources. Reader votes don’t move them.
What the scores mean
G is the gain side, out of 100. It goes up when the gain is specific, when it measurably helped people and when the sources behind it are strong.
P is the problem side, out of 100. It goes up when the problem is specific, when it measurably hurt people and when the sources behind it are strong.
In brief
The effect worked through broken expectations. Highly humanlike chatbots led consumers to expect socially appropriate and competent conversation, and a process failure violated those expectations more sharply.
After a process failure, an apology reduced customer aggression more than telling customers they were dealing with an AI, and the gap was larger for highly humanlike chatbots. The authors present this as guidance for firms designing chatbot recovery strategies.
Main points
- The studies combined three pretests (N = 30, 120 and 240) with two scenario experiments (N = 346 and N = 778).
- Consumers compared high- and low-anthropomorphic chatbots in service settings.
- Humanlike chatbots intensified customer aggression under process failure, where the failure concerned the interaction itself.
- The effect was mediated by disconfirmation of anticipated expectations of socially appropriate, competent interaction.
- After process failure, an apology worked better than acknowledging the chatbot's AI nature, particularly for highly humanlike chatbots.
The full story
The research separated process failure, where the interaction itself goes wrong, from outcome failure. Humanlike design raised customer aggression in the process case, which the authors tie to expectations being violated.
Two recovery responses were tested: an apology and an acknowledgment that the chatbot is an AI. The apology reduced aggression more after process failure, with the largest benefit for highly humanlike chatbots.
The authors say the work clarifies when humanlike chatbots help or harm consumers in service encounters and offers firms practical guidance on designing recovery strategies.
What this doesn’t fix
The findings come from scenario-based experiments in which consumers evaluated described or simulated interactions, not from observed real-world customer behaviour.
Sources
1 on file- Journal of Consumer Behaviourdoi:10.1002/cb.70240
Talk it over
This record found a gain and a problem in the same place. Which matters more to you, and who might answer differently?
Humanlike chatbots anger customers more when service goes wrong, experiments findhttps://truvace.com/item/mitigating-chatbot-service-failures-the-roles-of-anthropomorphism-acknowledgment
This record found a gain and a problem in the same place. Which matters more to you, and who might answer differently?
The gain In two scenario experiments (346 and 778 consumers), an apology after a process failure reduced customer aggression more than acknowledging the chatbot's AI nature, especially for highly humanlike chatbots.
The problem In two scenario-based experiments (346 and 778 consumers), highly humanlike chatbots intensified customer aggression after a process failure because they violated stronger expectations of polite, competent interaction.
Answer in the threadHow do we know? What can this kind of evidence show, and what can’t it?
The evidence This rests on a controlled comparison with a sample of 1,124, from Journal of Consumer Behaviour.
Answer in the threadWhat does it leave open? What else would you want to know before acting on it?
The record’s own caveat The findings come from scenario-based experiments in which consumers evaluated described or simulated interactions, not from observed real-world customer behaviour.
Answer in the threadHave you seen AI used this way where you study, work or live? Did it go the same way?
Answer in the thread
The questions are the same on every record. The details under them are taken from this record as written.
What readers say