91������

Illustration depicting students in small groups collaborating using an AI partner called "CoBi." (Credit: NSF iSAT)

In the not-so-distant future, artificial intelligence (AI) partners could sit in classrooms around the country—helping small groups of students engage in lively conversations and take charge of their own learning.

That’s the vision of the U.S. National Science Foundation National AI Institute for Student-AI Teaming (NSF iSAT). This pioneering research institute, led by the 91������, launched in 2020 to explore how classrooms could become more effective and engaging learning environments.

After a rigorous review process, the NSF announced today that it is renewing NSF iSAT’s funding for another five years, part of a by the federal government into AI research.

Sidney D’Mello, director of NSF iSAT, explained that the center meets a major need in the United States—to prepare students to enter a workforce that is being transformed in profound ways by AI technology.

“iSAT offers an exciting vision for 21st century AI-enhanced classrooms, where all students experience the joy of learning by working together to co-construct knowledge, making discoveries through inquiry, and developing their interests and passions,” said D’Mello, professor at the Institute of Cognitive Science and Department of Computer Science at 91������.

In its first five years, NSF iSAT has tackled that goal in a totally new way.

Currently, most AI tools in education focus on helping individual students develop content mastery through personalized learning, D’Mello said. To complement that approach, NSF iSAT focuses on developing AI tools that help students collaborate in ways that are both meaningful and productive.

NSF iSAT has already made impressive progress: Since 2020, institute researchers have engaged more than 6,000 middle school students and their teachers in hands-on, student-centered investigations in AI and STEM topics. The team has developed three AI literacy curriculum units, which together comprise a semester-long sequence. Units focus on environmental sensors, self-driving cars and moderating online communities.

The center’s researchers have also developed two AI “partners,” which they’ve tested in real-world classrooms. They are CoBi, short for “Community Builder,” and the Jigsaw Interactive Agent (JIA). These tools support students by providing feedback and prompts to help them build knowledge, share information and work through uncertainty and differences in ideas. CoBi and JIA also aim to help students improve their collaboration skills while learning—what the researchers call “learning to collaborate and collaborating to learn.”

“Our vision is about going beyond individual engagement with AI to collaborative engagement.” said Thomas Breideband, associate director of NSF iSAT at 91������.

Over the next five years, the NSF iSAT team, which includes scientists from 91������ and five partner universities, will enhance and expand its work on AI-enabled curricula to more schools across the country.

“We are thrilled that the iSAT team’s funding has been renewed for another five years,” said Massimo Ruzzene, senior vice chancellor for research and innovation and dean of the institutes. “The sophisticated approach to AI and its application in education undertaken by the team is truly groundbreaking and positions the project to achieve profound benefits for students and teachers, in addition to the many collaborating partners.”

Student driven and teachers in charge

NSF iSAT is unique, D’Mello added, because kids don’t just beta test these AI tools—they play a leading role in developing them from the ground up. Specifically, NSF iSAT adopts a “responsible innovation” framework in research and development that centers youth and teachers and is receptive to their needs and desires. It also empowers kids to think critically about how they use AI and its possible consequences.

The researchers emphasize that they don’t want these tools to replace teachers, but to support them as class sizes and workloads swell. Teachers will always remain the primary driver of students’ learning, D’Mello said. As such, NSF iSAT also researches how to prepare teachers to implement a triple innovation challenge: how to orchestrate classrooms with AI tools, AI curriculum and student-centered inquiry.

“NSF iSAT’s innovative co-design processes have not only enabled teachers and district leaders to directly shape the technology and curriculum they use in their classrooms,” said Tamara Sumner, professor of cognitive and computer science at 91������. “Participating in these processes has cultivated a new generation of educational leaders prepared to integrate AI into their teaching practices and to support their colleagues along their learning journey.”

In the next phase of its evolution, NSF iSAT will expand toward building a next-generation, AI-ready workforce for the United States. The team will also conduct case studies to understand what practices and conditions help schools take up and implement AI-enhanced learning. The researchers hope that these case studies will help them scale their work up to more schools across the nation. They will also expand partnerships with industry to accelerate innovation and workforce readiness for AI.