IBM’s famous Watson computing system—which defeated Jeopardy champ Ken Jennings in 2011—is coming to education, if not quite the classroom. As part of a new IBM philanthropic initiative, the supercomputer is helping to power a searchable database of open educational math resources designed for teachers in grades K-5.
Today marks the first time the new tool, called Teacher Advisor With Watson 1.0, is open to the public after a lengthy beta testing period that sought input from state education commissioners, teachers unions, school board associations and more than 1,000 teachers.
“We wanted to build and design something for teachers by teachers, with the best information and the best technology available,” says Stan Litow, the President Emeritus of the IBM Foundation and a former deputy chancellor for New York City Department of Education.
The IBM Foundation has been flirting with ideas to apply Watson technology in education for a while, without knowing exactly what it wanted to do with it. The tech giant began last year by pulling in more than 100 top-level education leaders for a daylong event demoing the tech. From that focus group they narrowed the list of potential applications to professional development tools and, eventually, settled on a searchable database exclusively for elementary school math.
“When we went back to the advisory committee, we asked, ‘Where would you start if you were looking to make the largest impact?’" Litow says. “The consensus was: start with math at the elementary level, because those teachers are usually licensed as elementary teachers—they may not have strong subject-level expertise. If you could focus in on math, that would be the moonshot.”
The result is a smart search engine that uses Watson’s ability to parse natural language and make recommendations with the aim of accurately matching what teachers are really looking for.
Populating the search engine is a collection of more than 1,000 OERs—from sources such as Achieve, UnboundED and statewide orgs like EngageNY—hand-selected by math experts assisting the program. When teachers search a particular concept, such as fractions or place value, they get a targeted lesson, standard, recommended activity and strategy. From there they can adjust variables like grade level while keeping search terms the same, which may assist teachers in classrooms where students have different math proficiencies.
The real hook, however, is Watson's continuously evolving AI. Theoretically, the search tool should improve and refine itself based on data from users, potentially providing more relevant results as time passes.
Within the OER content bank, there are print materials as well as diagrams and even a video player, which takes advantage of Watson tech that can pinpoint the specific parts of any given video relevant to the search query and cue up just that part.
Since launching the beta late last year, IBM has sought input from a cast of thousands, both online and at in-person workshops. Christine Manna, a math coach for the Waterford Township School District in New Jersey, attended a demo workshop back in January and, along with other educators, delivered some blunt feedback. The layout wasn’t user-friendly—it was clunky and tough to navigate when time was of the essence. From all accounts, IBM listened.
“Thankfully they made the changes and really heard what we had to say,” says Manna. “Coming back and revisiting this tool later on in the school year, I saw great improvement.”
Now, it’s possible for her teachers to differentiate lessons on the fly, punching in only a few keywords to get started. “When you go and research in a specific area—for example, equivalent fractions—you can look at the different designs of each lesson,” Manna says. “You can look at the wording and see very quickly if it’s higher [difficulty] or more for students who struggle. It cuts down all of the work for you. I think that is most appreciated.”