The new science of teaming

Artificial intelligence in classrooms could add up to real advances in education

Take a seat in the classroom of tomorrow—where intelligent computers work side-by-side with groups of students to support their engagement in meaningful and productive learning experiences designed by their teachers.

That’s the vision of a new $20 million research collaboration led by CU Boulder called the U.S. National Science Foundation (NSF) AI Institute for Student-AI Teaming. The effort is exploring the role artificial intelligence could play in the future of education and workforce development, especially in providing new learning opportunities for students from historically underrepresented populations.

Sidney D’Mello, an associate professor in the Institute of Cognitive Science (ICS) and the Department of Computer Science, leads the new institute. The five-year project will bring together a team of researchers from nine universities in close collaboration with two public school districts and many private companies and community leaders. It will also tap researchers from across the CU Boulder campus.

“We aim to advance a new science of teaming,” D’Mello said. “We have a lot of knowledge of what makes effective human-human teams. The next phase is understanding what underlies effective human-agent teams. In our case, that means students, AI and teachers working together.”

The project team hopes that its work will also inspire kids from Colorado and beyond to get excited about topics like AI and computer science.

“This center aligns with our vision of producing research that quickly translates into meaningful societal impact,” said Keith Molenaar, interim dean of the College of Engineering and Applied Science.

D’Mello explained that his team’s work touches on a problem that’s familiar to anyone who’s set foot inside a K–12 classroom recently.

“Researchers and educators have talked about how important collaboration is to effective learning for a long time,” D’Mello said. “It’s just really hard to do that in a classroom because the teacher can’t be omnipresent.”

Imagine an intelligent agent that could follow what groups of students are talking about, then ask questions or provide feedback to enhance their learning. At the same time, the agent works with teachers, helping them orchestrate more effective classroom interactions, such as by providing summaries of the small group discussions.

To make such a reality happen, the new institute focuses on three main challenges: In the first, researchers work to develop new advances in the fundamental science of how machines process human language, gestures and emotions.

“Our AI partner needs to engage seamlessly with multi-party, studentled conversations in noisy classroom settings,” said Martha Palmer, coprincipal investigator on the project and professor in the Department of Linguistics. “This poses many novel challenges for speech recognition as well as dialogue understanding, making iSAT the most daunting and the most exciting project any of us have ever been involved in.

Next, the team strives to better understand how students, AI and teachers can collaborate effectively in both classrooms and remote learning contexts. Last, researchers will go to classrooms in Denver Public Schools and other school partners to ask students and teachers to help them think up new ideas for technologies. 

The idea of intelligent machines in schools might make some parents uneasy. The project, however, draws on an approach to research called “responsible innovation” to develop tools that schools actually need.

“Community members must be included from the very beginning when it comes to designing and developing technology that will be deployed in schools—this includes involving students, teachers, parents and other community leaders,” said Tamara Sumner, who is part of the new institute’s leadership team and the director of ICS, as well as a professor of computer science.

William Penuel, a member of the leadership team and professor in the School of Education, added that AI is only going to become a bigger part of our world in the next few decades—so kids need to know how it works and how it can work responsibly.

“Students need to understand how AI functions in the world now, including its potential role in building a ‘surveillance economy,’ and how it can help communities design together for a more just future,” he said.