Why We're Building Good Robot Teachers

Aug 30 2017

Shlomy Kattan, PhD

In a recent New York Times Sunday Review op-ed, David DeSteno, Cynthia Breazeal and Paul Harris, professors of psychology, media arts and education at, respectively, NYU, MIT and Harvard, responded to a spate of findings that education technology programs have little to no evidence of improving academic performance.

The problem, the authors write, stems in part from developers' relative neglect for the importance of social interaction in learning. "The human brain is an amazing information processor," they write, "but it evolved to take in, analyze and store information in a specific way: through social interaction . . . [The mind relies] on and incorporate[s] social cues to facilitate learning."

The researchers summarized two studies they conducted with young children. In one, a robot told 4- to 7-year-old children a story with either flat or expressive affect. Children's engagement, attention and both short-term and long-term retention of new vocabulary and plot were evaluated using eye scanning software and by having the children repeat the story immediately after hearing it, and again after 4 to 6 weeks. Not surprisingly, the children who heard the story from the robot showing expressive affect were more engaged and retained more of the new vocabulary and plot, especially longer term.

In a second article, reporting on a study conducted with preschoolers, the researchers measured how much children trusted information they received about a made-up animal from two robots based on whether or not they behaved in socially contingent ways. Although the two robots conveyed the same information, when the novel animal was shown to the children at the end, 82 percent of the kids chose to seek further information about the animal from the robot that responded to children's behavior and utterances as opposed to its partner.

Taken together, these studies suggest that, as the authors write, "if we want to use technology to help people learn, we have to provide information in the way the human mind evolved to receive it." That is, socially.

We could not agree more. Learning is a social process, not a solitary one. Both of our XPRIZE competitions addressing learning posit as much.

The Global Learning XPRIZE challenges teams to develop open-source scalable software that will enable children in developing countries to learn basic reading, writing and arithmetic alone and with others. Likewise, among the requirements for the Barbara Bush Foundation Adult Literacy XPRIZE, which challenges teams to develop mobile software for low-literacy adults, is to include a social learning component so that learners can tap into the power of their communities. These requirements exist precisely because learning happens best socially.

However, the authors neglect one important value of education technology that goes beyond its measurable relative effectiveness: access.

More than one billion people around the world lack basic skills. Nearly 300 million children, mostly in rural areas of developing countries, lack access to a meaningful and effective educational experience, and more than 780 million adults lack basic literacy and numeracy. The vast majority of these children are girls, and two-thirds of these adults are women.

To reach the goal of providing universal basic education, the UN estimates that the world needs 69 million more teachers, a staggering figure. Technology can go a long way in closing that gap.

Some promising examples already exist. The Rumie Initiative, for instance, is a non-profit that makes access to free digital education possible for underserved communities worldwide by distributing low-cost tablets with free and open-source educational software. Likewise, Plan Ceibal in Uruguay has provided internet access and tablets to every elementary school student in the country, enabling children to access apps like codeSpark Academy, which teaches the ABC's of computer science, a subject matter these kids would otherwise not be able to learn. And at a time when we are facing the worst refugee crisis since WWII, NetHope's No Lost Generation Task Force is working with a consortium of non-profits, for-profits and government agencies to provide basic access to education and internet for the more than 30 million refugee children.

Access for low-literacy adults often means something different in industrialized countries. Adult learners lead busy lives and may not even be aware of their low-literacy status. Attending class between work and home isn't always a feasible option. And even when it is, the total capacity for adult education in a country like the United States—where 36 million adults lack these basic skills—can address the needs of only 5 percent of the adult learner population. Providing high-quality educational contentfor adult learners that is accessible anytime and anywhere can begin to address that gap.

Yes, socially responsive and appropriately behaved robots can be more effective educators than products available on the market today. But for the hundreds of millions of learners around the globe who need basic support, an app can open up a world of opportunity.

Shlomy Kattan, PhD