In a study of human-chatbot interaction from ResearchGate, researchers found that participants who interacted with a bot designed to be as “human” as possible reacted negatively, experiencing a sense of discomfort from the interaction. So, some bot designers have concluded that it’s essential to keep their creations from promising too much. “When customers think that a bot is human, or capable of human-level interaction, they will often speak to it in a much more roundabout way,” Pranay Jain, co-founder and CEO of Enterprise Bot, a company that builds bots for companies, said in an email interview. “This inevitably creates a communication issue between the bot and the person, and because their expectations are higher, they are left more disappointed.”
Set Boundaries for Your Bot
Some research has found that users prefer talking to people instead of bots. But for those who do like communicating with bots, there are ways to make the interaction more appealing. AI today is brilliant, but it isn’t perfect, Jain said, so chatbots need to set boundaries from the start of a conversation regarding capabilities. “The honest truth is no one wakes up in the morning and thinks, ‘Hey, I’d love to talk to a chatbot today,’” Jain said. “All they want is a solution to their problem. When it’s clear that the conversational AI isn’t human, it helps adjust the users’ expectations and change their behavior towards the bot.” Knowing when and where to apply conversational AI is critical for companies, Joseph Ansanelli, co-founder and CEO of Gladly, a customer service software firm, said in an email interview. Information that simply requires a lookup is one area where bots shine. “But for nuanced conversations like questions about the fit of a certain product or seat selection for an airline—those are bad fits for conversational AI and should be routed to a human who can interpret customer intent and provide personalized answers,” Ansanelli added. Not everyone believes that modest bots are the future. Many in the software industry say that chatbots should be as human-seeming as possible. “Natural-sounding virtual agent voices and dialog patterns make task completion more enjoyable and efficient for most people,” Evan Macmillan, the CEO of Gridspace, a company that makes software for call centers, said in an email interview.
Smarter Is Better
Some experts say bot appeal simply comes down to smarts. “People love chatbots but hate dumb chatbots,” chatbot developer Stephen Blum, the chief technology officer of PubNub, said in an email interview. “It’s easy to spin up a rule-based chatbot, one that has pre-made responses triggered by pre-determined inquiries, but when it comes to engagement beyond simple question-and-answer, you have to build intelligence into your chatbot.” One way to keep bot-human relations friendly is by incorporating nuance into the conversation, Michael Ringman, chief information officer at TELUS International, a company that advises on digital customer experience, said in an email interview. He suggests incorporating cultural nuances into bots. Companies should “employ local experts in the different countries and regions where your customers live to ensure cultural considerations and regional expressions are included in the bot’s language library,” Ringman added. Speed wins over users when it comes to chatbots, it seems. “The modern consumer has been groomed to expect nearly instant gratification,” Evan Chen, the co-founder and CEO of Akia, an AI-powered hotel guest communication platform, said in an email interview. . “When at a hotel, having questions like ‘What’s what Wi-Fi password?’ answered immediately is far more convenient than calling or waiting for a reply (even if it is 60 seconds).”
