Should companies using AI chatbots be transparent about the involvement of human employees working behind the scenes? Do customer interactions change when human involvement is disclosed? If so, how and why? Our recent paper published in Information Systems Research written by Ulrich Gnewuch, Stefan Morana, Oliver Hinz, Ralf Kellner, and Alexander Maedche offers some answers to these questions. 

--- 

Summary of Key Findings 

Our findings suggest that disclosing human involvement before or during an interaction leads customers to adopt a more human-oriented communication style. Rather than using simple keyword-style queries, customers tend to use longer, more complex, and more natural sentences when human involvement is disclosed (vs. not disclosed). This effect is driven by customers’ impression management concerns. That is, customers are more concerned about making a good impression when they know that there are human employees who step in if the chatbot is unable to respond. Ultimately, the more human-oriented communication style increases employee workload because fewer customer requests can be handled automatically by the chatbot and therefore must be delegated to a human. These findings are important because they help us understand how customers respond to human-AI hybrids and reveal how transparency about human involvement can place additional workload on employees working in tandem with AI. 

Link to Paper 

https://pubsonline.informs.org/doi/10.1287/isre.2022.0152