Should Chatbots Disclose That They Are Chatbots?

Once a technology on the cutting edge, chatbots are increasingly common. Most of us encounter them regularly: when we interact with our phones or computers, when we call a support number, or when we check into a building through a security checkpoint. The benefits of chatbots are clear: they can provide customers with round-the-clock support and help distribute calls and contacts to the right human agents.

But do humans like interacting with them? Many of us have had experiences with poorly designed chatbots, who “didn’t quite get” what we said…repeatedly. New research has borne out this suspicion. A study conducted by Göttingen University found that "chatbot disclosure has a negative indirect effect on customer retention through mitigated trust for services with high criticality."

The “high criticality” part is important. A Genesys (News - Alert) survey from 2019 found that 73 percent of respondents were open to dealing with a voice/chatbot. When issues were more complex, however, only 21 percent felt comfortable dealing with bots. In other words, we don’t mind carrying out low level tasks such as call routing or basic “what are your hours?” questions with chatbots, but when we have a more complex problem, we want a human being.

Another pain point for customers is chatbots that pretend to be human beings, according to a recent article by digital marketing company iQuanti. With conversational AI, chatbots are capable of understanding topics, recalling data about a specific user between conversations, and learning from continuing interactions with customers, similarly to a real customer support agent. But do we want them to?

“While customers have daily AI-enabled interactions with organizations, trust remains to be an area for improvement,” according to the company. “Customers want more human-like interactions, but still prefer to speak to a live representative most of the time. More than 70 percent of organizations are actively trying to make their AI interactions more human-like, but this raises the question of whether chatbots should "pretend" to be real, or if it should be clearly disclosed the bot is indeed a bot.”

Experts recommend that companies design their chatbots to be transparent to customers (so customers know they’re talking to a bot) and ensuring there is an easy path to connect to a human agent if the customer wishes. 

Edited by Maurice Nagle