Was this forwarded to you by a friend? Sign up, and get your own copy of the news that matters sent to your inbox every week. Sign up for the On EdTech newsletter. Interested in additional analysis? Upgrade to the On EdTech+ newsletter.
The flawed University of Maryland report that was not really about AI tutoring could have had a more interesting approach if the researchers had stuck to their own advice.
The findings reported here thus concern the [GPT-4o–based virtual study assistant] VSA as configured and introduced in this study, rather than establishing the effects of more scaffolded AI tutoring.
If you read the report as a study on AI tutoring (which was the title and assumption behind the study), one question that jumps out is how anyone in their right mind would design an AI tutor that by default gives direct answers to students—essentially a university-sanctioned tool to bypass the instructor’s teaching process. If you read the report as product research on UMD’s VSA, however, it looks completely different.
What emerges is a well-intentioned course assistant chatbot where substantial development and documentation did not translate into sustained adoption or instructional integration. And it is not surprising that only 33 instructors signed up for the study, and just 15% of students in the treatment group even tried out the VSA.
What VSA Is
Subscribe to Premium to read the rest.
Become a paying subscriber of Premium to get access to the rest of this post.
Upgrade