Improving education systems and outcomes

If you want a glimpse into cutting-edge research on artificial intelligence in education, look to Jing Liu, director of the Center for Educational Data Science and Innovation at the University of Maryland. Two studies released in October offer a striking and enlightening contrast. One hinted that a little AI-powered math practice might help struggling middle schoolers improve their skills. The other found that an AI assistant undermined college students’ engagement and hurt their grades.

Liu wanted the studies to be released together, he told me, to show that AI in education is neither “all good” nor “all bad.” The field is still figuring out how to design tools that improve learning over time. And that means studying not just the technology, but also how students respond to it — including in ways its creators never intended.

A large University of Maryland trial has produced a striking result for universities rolling AI into teaching: students offered a course-integrated AI tutor performed worse in matched classes and engaged substantially less with existing course materials.

The randomized study involved 2,379 undergraduate students and 30 instructors across multiple disciplines during fall 2025.

In the clearest comparison, between sections of the same course, students given access to the AI tutor finished around four percentage points lower in their final grades, equivalent to 0.37 standard deviations.

Concerns that suspensions disproportionately affect students of color have led many school districts to prioritize other forms of school discipline, including detention, loss of privileges, and restorative or behavioral interventions and supports. A new study by Youngsun Lee and Jing Liu at the University of Maryland, College Park, and Emily K. Penner at the University of California, Irvine finds that the strategy hasn’t closed at least one large district’s racial and ethnic discipline gaps.

Jing Liu, founding director of the Center for Educational Data Science Innovation and an associate professor in education policy at the University of Maryland, said issuing these guidelines are a good starting point, but it may be challenging to keep up as the technology rapidly develops.

“AI is very different compared to prior technology breakthroughs [because] they are evolving so fast and they can do things that are directly different this month compared to last month,” Liu said. “[The guidelines] cannot be a static thing. It was to be very adaptive.”

Accelerate today named Eedi and Edlight among its latest round of grantees under the Call for Effective Technology (CET) program, adding two new AI-enabled learning tools to EDSI's evaluation portfolio. With this award, EDSI will evaluate both platforms in public school classrooms during the 2026-27 school year.

This project uses AI-supported analysis of recorded classroom interactions to uncover the student skills embedded in real‑time instructional exchanges and assess how well these competencies—spanning both core mathematical reasoning tied to upward mobility and more durable skills such as agency, vocabulary use, and self‑management where disparities often emerge—serve as early precursors of long‑run mobility.