If you’ve visited an online retail store, you’ve likely encountered a chatbot before. It’s the small message that pops up in your screen’s bottom corner, saying something along the lines of, “Hi there! Thanks for visiting our site. Can we help you look for something?”
Elizabeth Han, a Desautels Information Systems professor, investigates these AI chatbots and their use in the modern world. She recently published an article in //Information Systems Research// looking into whether it is beneficial for AI chatbots working in customer service to express, or display, positive emotions.
Existing research suggests that when human customer service representatives express positive emotions, customers feel like the quality of service is higher. Psychologists theorize this to be a result of “emotional contagion”—when a person sees someone else feeling an emotion and they begin to feel it as well.
The question Han and her team set out to answer is if this effect translates to exchanges with chatbots. They hypothesized that while emotional contagion may still have an effect, it could be negated by customers’ feelings of discomfort toward a machine expressing human emotions.
To answer this question, Han and her colleagues asked participants to complete online tasks using virtual chatbots, like requesting an exchange for a textbook they had previously ordered. Some participants interacted with chatbots that were programmed to express positive emotions, while others interacted with ones that were more robotic.
The emotion-expressing chatbots used lines like “I can help you with that, and I am excited to do so!” while the straightforward chatbots kept it to a simple “I can help you with that.”
After participants completed these interactions, they rated the quality of the customer service experience. Han and her colleagues found that customers who interacted with positive-emotion chatbots did not experience the increase in perceived service quality that you would have expected had they been interacting with human agents.
This new information presents businesses with a tricky question: Save money by replacing customer service employees with chatbots, or provide a better customer experience by continuing to use human agents?
In an interview with The McGill Tribune, Han explained how the switch to chatbots increases company efficiency.
“Essentially, these service employees are doing emotional labour by interacting with perhaps annoying customers, but chatbots are not really affected by that,” Han said. “The chatbot doesn’t feel anything, and they can just make whatever appropriate response.”
Given how much more efficient chatbots are, some companies may be tempted to simply not tell customers that they are interacting with a chatbot instead of a human.
“There is research that has been done that has looked into the effects of disclosing the chatbot’s identity, and it actually reduces those business outcomes. There is a tradeoff between better business outcomes versus keeping the line of ethical and legal codes,” Han explained.
This then becomes an ethical and legal question: Do customers have a right to know when they are speaking to a computer? And would companies honour this right if it meant a loss in profits?
“I think there are many regulations coming up regarding disclosure so eventually companies will move more and more towards disclosing the identities of those chatbots,” Han said.
The field of AI is rapidly developing, and thinking about how these AI chatbots have been created and will continue to evolve raises many difficult questions. It might even make you wonder the next time you’re browsing an online store and get the pop-up notification to chat: Are you talking to a human or a machine?