The Lab

Tucked into the third floor of the Whitfield Science Center, the Adaptive Learning Lab doesn't look like the future of education. A dozen graduate students sit at workstations, surrounded by whiteboards covered in diagrams and sticky notes. But the work happening here has drawn attention from universities, ed-tech companies, and policymakers across the country.

Led by Dr. Priya Sharma, Associate Professor of Computer Science and Education, the lab is developing AI tools that help instructors identify when students are struggling — and intervene before they fall behind.

The Problem

"The traditional model is reactive," Dr. Sharma explains. "A student fails a midterm, and then we notice. By that point, they've been lost for weeks. We wanted to flip that."

The team's system, called EduPulse, analyzes patterns across learning management systems, assignment submissions, and discussion forum activity. It doesn't read student work — instead, it looks at behavioral signals: submission timing, engagement frequency, help-seeking patterns.

When the system detects a student whose patterns diverge from their own baseline, it alerts the instructor with a summary and suggested actions. No grades. No surveillance. Just a gentle flag that says, "This student might need a conversation."

Early Results

A pilot across three departments last semester showed promising outcomes:

• 40% reduction in students scoring below C in pilot courses
• 3x increase in students visiting office hours after instructor outreach
• 92% of participating faculty said they would continue using the tool

"It caught things I wouldn't have noticed until the final," said Dr. Robert Kim, who teaches Introduction to Economics. "One student had stopped participating in discussion boards three weeks in. I reached out, and it turned out she'd taken on a second job and was overwhelmed. We worked out a plan. She finished the semester with a B+."

Ethics First

The team has been deliberate about building ethical guardrails from the start. EduPulse does not use demographic data. It does not predict grades. It does not make decisions — it surfaces signals for humans to act on.

The project has a dedicated ethics review board that includes students, faculty, and an external privacy advocate. All data is anonymized in research publications, and students can opt out at any time with no impact on their grades.

"We're not building a system that replaces instructor judgment," Dr. Sharma emphasizes. "We're building one that gives instructors better information, earlier."

What's Next

The team is preparing to expand the pilot to all undergraduate courses in the College of Arts & Sciences this fall. They've also received a $2.4 million grant from the National Science Foundation to study long-term outcomes and develop an open-source version that other universities can adopt.

For Decoupled University, the project represents exactly the kind of research the institution was founded to support: practical, ethical, and focused on making education better for everyone.