Hospitality workers trusted poor advice more from AI than from human experts, experiments find

The finding points to automation bias in workplaces that use AI: employees may defer to a machine source even when the content is poor. The study measured trust, perceived effectiveness and behavioral intention, so the result concerns what employees said they would do in scenarios.

The authors report that critical thinking skills and cognition-based trust may help reduce the effect. They say both information quality and the agent delivering it determine employees' psychological state and behavior where AI is integrated, and they offer guidance for managers who want to introduce AI ethically.

The study, published in the International Journal of Contemporary Hospitality Management, set out to test the interactive effect of information quality and agent on hospitality employees. Across five scenario-based survey experiments, the researchers varied whether information was high or low quality and whether it came from an AI expert or a human expert.

Employees' responses covered trust, perceived effectiveness and behavioral intention. The central result is that weak information gained an advantage in trust and uptake when an AI expert, not a human expert, was its source.

The authors say critical thinking skills and cognition-based trust may help counter this effect. They frame the work as a contribution to automation bias and source credibility theory, and as practical guidance for managers who want to implement AI technologies ethically.