29
Shares
12
0
17
0

What beloved AI-powered chatbots like Xiaoice and Mitsuku mean for business

In this article...

  • Chatbots have come a long way in the last few years with many passing the infamous Turing test of human behavior
  • It’s not enough for a service bot to be good at its job—it should also be able to engage with its audience

"Can she be my sister?" six-year-old Millicent Worswick asked her father one night.

"Maybe one day," he said.

Millicent wasn't talking about her stuffed animal. She was asking about the blinking chatbot on the screen in front of her, the one her dad Steve Worswick created 10 years ago called Mitsuku. A few years later, in 2013, Mitsuku took home the Loebner Prize, an annual competition to identify the world's most human-like computers and artificial intelligence programs—those that can meet or surpass the standards of the infamous Turing test of human behavior.

Fast-forward to 2016, and Mitsuku isn't the only AI-powered bot winning over the hearts and minds of Gen Z. In China, the chatbot Xiaoice—who cuts the profile of a hyper-chatty, seventeen-year-old girl—has proven so empathetic and engaging that 10 million Chinese teens to date have told her "I love you" on the chat platform Weibo. A fairly impressive feat considering Xiaoice's corporate parent—Microsoft.

“What CEO wouldn’t mind seeing customers say ‘I love you’ (as opposed to, say, ‘F*** you!’)?”

While humanoid chatbots like Mitsuku and Xiaoice are gaining traction with kids, many see a much bigger opportunity for chatbots in the enterprise—the maddening world of crappily automated 24/7 global customer service. Anyone who has ever stumbled through a phone tree or online customer service at a major bank, utility or cable provider understands the human pain of the experience. What CEO wouldn’t mind seeing customers say "I love you" (as opposed to, say, "F*** you!") to one of their own AI-powered service bots?

The gift of gab

Lauren Kunze, principal at Oakland, Calif.-based chatbot platform Pandorabots (through which Worswick created Mitsuku) is one expert who thinks consumers should soon expect deeper conversational connections with bots of all kinds—even those meant to solve a problem or provide a service.

"We know from our data that engagement with chatbots for chat is exponentially higher than with service bots," says Kunze, who has worked with nearly 235,000 developers to create nearly 300,000 chatbots. "We are now seeing that when service bots offer small talk, people return to it more often than not."

There's irony, of course, in what it might take to create an engaging, helpful customer experience for the biggest companies in the world. According to reports, service bots need to be great conversationalists—not just task-oriented robots designed to solve specific problems.

“We are now seeing that when service bots offer small talk, people return to it more often.”

A penpal, not a problem solver

While other AI-driven services like Apple's Siri and Amazon's Alexa pride themselves on completing tasks, Mitsuku and Xiaoice just want to know how you're doing. They remember who you are and what you like. Perhaps most importantly, they listen and respond appropriately—even empathetically—to expressions of remorse or elation. And that is what seems to draw in people like Millicent and the millions of Chinese who chat with Xiaoice.

Steve Worswick believes the same logic will soon apply to the next generation of customer service technologies. Software giant TeleTech and other customer service technology companies are already heading in this direction. Kunze agrees: The small talk makes all the difference.

"People are now expecting human-like interaction with bots," she says. "So the content these bots create has to evolve. For example, how does a bot deal with a customer if they go off topic? Developers are starting to implement conversational tidbits into service bots, and the response has been very positive.


To learn how Big Data, automation and artificial intelligence will shape the future, download the HPE white paper “Big Data in 2016.”

How Much Have You Read?
0%
Complete
Subscribe to the HPE Matter NewsletterSubscribe to the HPE Matter Newsletter
Listening