Illustration: Dale Edwin Murray

In the spring, if all goes according to plan, about 15 middle-school students in Westchester County, New York, will walk into an IBM laboratory where technology meets education. They’ll leave their textbooks behind and strap on virtual-reality headsets. They’ll check the settings and blink a few times to adjust their eyes. And then they’ll start a teacher-approved earth-sciences tutorial about gravity—by flying past Saturn and Jupiter to understand how gravitational pull affects the planets.

As they do this, the computer- and graphics-driven tutorial will not only be teaching but also learning how well the lesson is understood by each student. “The whole goal is to provide a more engaging learning experience that is orchestrated by the children,” says Satya V. Nitta, program director and master inventor in IBM Research’s Learning Sciences and Cognitive Computing Department.

The planetary fly-by is just one of the pilot programs IBM has undertaken as it works to use the predictive capability of cognitive computing to enhance the way kids learn. Its ambition is to develop new technology for education based on neuroscience and advanced data analytics, and the goal of that technology is twofold: to facilitate learning and to use data collected from the students and elsewhere to give teachers insight into when their kids are learning and when they’re struggling. IBM calls the desired result—in effect a tool teachers can use to learn their students—the Intelligent Tutoring System.

The 3-D gravity lesson may seem like just a fun new way for kids to learn, but in fact it’s a project that hits both of IBM’s objectives: speeding students’ learning with a virtual-reality experience and giving teachers feedback they can use to improve each student’s success in school long after the lesson is complete.

In comparison, think of a 14-year-old sitting at a desk listening to a lecture about gravity. The teacher may not realize a student’s eyes have glazed over while hearing about the Kerr solution for rotating massive objects, and even if the teacher does notice, she may not have another way to explain the concept. But the virtual-reality system will be able to tell if the class seems confused or has stopped paying attention. The computer won’t replace the teacher’s judgment, but thanks to cognitive computing, it will have the ability to predict, based on all its interactions with students, when another kind of lesson might work better.

“If it appears they’re not getting it,” Nitta says, “the computer will suggest another experience, maybe a dialogue.”

While this type of predictive system may sound futuristic, it’s actually more like “Back to the Future.” Virtual reality is new, but IBM is basing the system design on research into which types of teaching have worked best throughout human history. The teaching inside that virtual-reality headset actually mimics the way kids have learned new skills from time immemorial.

“In historic times, if you were going to be a blacksmith, you learned the trade through a whole bunch of experiential skills,” Nitta says. “You watched somebody, you tried it yourself, you learned it.” The lesson worked because the body learned and the brain remembered the experience, not because you read it or somebody wrote it on a chalkboard. Like the blacksmith’s apprentice, a kid flying past a planet in a virtual environment is learning by doing.

“What we’re really saying is that we want to bring in as much of a real experience for the body as we can,” Nitta says.

IBM believes the concept can work in the classroom on other levels too. For example, another pilot program is designed to familiarize students with topographic maps. Teachers told IBM that topographic maps have long been hard for kids to learn because they’re a 3-D subject being taught in a 2-D way. So IBM figured out a way to make the lesson more interactive and more physical—by using tablets to let kids create 3-D augmented-reality holograms that hover before their eyes.  

“They can create a mountain that they can interact with,” Nitta says. “They can move it up and down, take a slice, move it around, whatever they want.”

Like student data input from the virtual-reality lesson, the tablets record how students progress through the lesson, including what was fumbled, what was mastered, and when. To protect the students’ confidentiality, the collected data is for teacher and school use only. Whatever data may be used otherwise remains strictly anonymous.

IBM’s ambitions for the uses of cognitive computing in education are now being tested at Georgia’s Gwinnett County Public Schools. In Gwinnett’s K-12 classrooms, multiple iPads and other data-collection tools are constantly amassing information about the district’s more than 170,000 students, and IBM is making those tools interoperable, so that all of their data can be compiled and analyzed as feedback for teachers.

IBM and the Gwinnett County schools have already learned that the more data that can be collected—and made sense of—regarding the learning that goes on in their classrooms, the more effective the learning experience can be. In fact, it can be customized down to the individual student, even within a school district as large as Gwinnett’s. The trick is finding a way to connect all the data that exists—having the attendance-keeping system talk to the grade-keeping system, which talks to the SAT-score system, which talks to the iPad lesson, which, someday, talks to the in-class virtual-reality system. Once that interoperability is achieved, cognitive computers turn all the information into predictive suggestions about exactly what help a student will need, and when.

In Georgia, this idea goes far beyond a single lesson about gravity or maps. The Gwinnett school district calls one result of this process an early warning system. IBM’s technology combines previously disparate sets of data about attendance, behavior, and course performance into a simple “red flag” warning for teachers who may not immediately see the signs or severity of a student’s problem.

While the Westchester program aims to help teachers identify students who are struggling with individual lessons, the Gwinnett project is aimed at predicting when a fourth-grader may be on a path to dropping out of high school years later.

Steve Flynt, chief strategy and performance officer for the Gwinnett County schools, says it’s already working. “We’ve gone back for an entire cohort of students from third grade to when they graduated and looked at the analytics of that group to see what came up in third or fourth grade that was predictive of them not being successful, maybe missing on-time graduation or not going to college,” he says. “We can go back and say—now with relative certainty—at this point in the year, X number of absences and X number of behavior issues along with these types of grades are very predictive, and it may be helpful to provide an intervention for that student. It’s an early warning system, the same as a business would use today.”

In fact, the systems IBM is using to create these educational tools include not only Watson (the cognitive computer that won Jeopardy!) but also systems now being used by some of the world’s largest businesses. To get millions of students’ information into the cloud while ensuring the students’ privacy and the data’s security, IBM is applying some of the same technology used by its global industry clients. Eventually, the child who suddenly understands how gravity affects Jupiter will benefit from the work IBM has done with everyone from Swiss bankers to New York hedge fund managers.

“IBM is probably the best company at managed data,” Nitta says. “Most of the world’s financial data goes through our systems. We know how to manage mass quantities of data at scale with privacy. In order to build all these things we’re talking about, you have to have this infrastructure, the foundation.”

It’s the kind of foundation that could be a launching pad for 13-year-olds with their sights set on Jupiter and beyond.