“What do the wheels of the bus do?”
Alex and Siri are nothing if not polite when they doesn’t understand a question. Adults know enough to try again and clearly annunciate. But what about kids? A new study by researchers at the University of Washington and published at the 17th Interaction Design and Children Conference, held in June in Trondheim, Norway, recorded 14 children aged 3 to 5 playing a Sesame Street Workshop game, “Cookie Monster’s Challenge” on a laboratory-issued tablet. But the researchers soon realized that children had their own challenges communicating with virtual assistants.
Children had difficulty trying to get a duck in the game to quack, and were told, “I’m sorry, I didn’t quite get that.” Some of the kids approached this problem differently from others. Of those 100 recordings, the children repeated their request 79% of the time. Some kept shouting “quack” while others tried to speak slower (“quaaack”). The children persisted 75% of the time without showing signs of frustration and only asked for help in 6 of the 100 recordings. The parents — not the children — decided to stop trying and declared a malfunction. Only then, did their children give up.
“Adults are good at recognizing what a child wants to say and filling in for the child,” said Alexis Hiniker, co-author of the study and an assistant professor at the University of Washington Information School. “A device could also be designed to engage in partial understanding, to help the child go one step further.” He said virtual assistants should ask follow-up questions to help kids as well as adults. “Instead of focusing on how to get the response completely right, how could we take a step toward a shared understanding?” (Amazon did not respond to request for comment.)
Don’t miss: Is Amazon’s Alexa smarter than Apple’s HomePod?
Smart speakers are creeping into U.S. homes. Earlier this year, Amazon released a $79.99 device, Echo Dot Kids Edition. Some 16% of Americans or 39 million people own a smart speaker, a recent survey by National Public Radio found. Of those, 11% had an Amazon Alex AMZN, +1.29% and 4% had Google Home GOOG, +0.27% That doesn’t include the millions of people who use Siri on their iPhones AAPL, +0.07% or voice-assistants on their Android devices. The most common tasks: playing music (60%), asking general questions (30%) and checking the weather (28%).
Language development occurs long before a child speaks and babies’ brains are equipped to understand more than one language and, Hiniker said, technology could play a valuable role in that. “Babies show a preference for the voices and even the language that they heard in utero,” Janet Werker, a language expert at the University of British Columbia, recently told Bold.expert, a blog about learning and development. “Listening, in early infancy, is essential for acquiring the syntax, sounds, and other aspects of language, enabling the child to understand and ultimately to speak.”
Hiniker says the persistent and patient reaction of the children in his experiment suggests there’s room for more sophisticated artificial intelligence to help children — especially those in multigenerational families where English is not the first language of members of the household. In fact, he has launched another study into how diverse, inter-generational families use smart speakers, and what communication needs are required. “They’re being billed as whole-home assistants,” he added. “Developers should be thinking about the whole family.”
Get a daily roundup of the top reads in personal finance delivered to your inbox. Subscribe to MarketWatch’s free Personal Finance Daily newsletter. Sign up here.