United StatesEducation2025 – ongoing

University of Maryland

Course-integrated Virtual Study Assistant

Maryland students offered an AI study assistant earned around 4 percentage points lower grades in matched course sections

In progressCompany's own claim

Problem

The university tested whether a course-integrated AI assistant could support undergraduate learning.

Approach

The Virtual Study Assistant used GPT-4o and retrieved instructor-selected course materials inside the learning platform. A randomized trial ran in fall 2025. Most instructors retained a mode that supplied direct explanations rather than guided tutoring.

Results

around 4 percentage pointslower final grades among students offered access in comparisons between sections of the same courseCompany's own claim[1]

around 15%of students offered the assistant used itCompany's own claim[1]

Takeaway

The trial measured the effect of offering access, including students who never used the assistant. It did not establish that AI tutoring generally reduced learning.

Sources

Students offered an AI tutor got lower grades and disengaged from course materials, University of Maryland trial finds
EdTech Innovation Hub2 Oct 2026News article

Source summaries

[1]EdTech Innovation Hub2 Oct 2026

The university trial found lower grades and less learning-platform activity among students offered the assistant. About 15% used it, limiting conclusions about active users.

Full details are in the original sources.

Added 1 Nov 2026 · Updated 1 Nov 2026