My brilliant brother, an engineer, and I were recently taking a walk with our dogs when he mentioned a concept that stuck with me. He suggested it would be valuable if instructors could personalize portions of general courses to help students find context in their specific areas of interest.
For example, an introductory course in AI could offer assessments and assignments aligned with a student’s major: a language learner would apply lecture knowledge to linguistics, while an electrical engineer would apply it to circuit design.
I’ve been chewing on this idea ever since. While I’m not a pedagogical researcher, it strikes me as a practical solution to a common hurdle in higher education: the “Gen Ed” disconnect.
The Challenge of One-Size-Fits-All
In a large lecture of 300 students, the core principles of a subject like AI are foundational, but they can often feel abstract. For many, these required courses feel like a “check-box” exercise rather than a toolkit for their future careers.
The idea my brother proposed is a form of Contextualized Teaching and Learning (CTL)—the concept that retention and engagement improve when students see a direct application of a principle to their own field.
A “Sandbox” for the Undecided
What really intrigued me was how this could help first-year students who are still exploring. Imagine if a student could:
-
Try a “Pre-Med” track for an AI assignment one week.
-
Switch to an “Architecture” focus the next just to see if it “clicks.”
-
Learn the same core logic as everyone else, but framed in a way that helps them explore potential careers in a low-stakes environment.
How the Tech Makes it Work
From my vantage point working with media and AI at Duke, the biggest hurdle has always been the instructor’s workload. No one can be an expert in 50+ different majors at once. However, a “modular” approach to information is starting to bridge that gap.
I’ve been working on an internal tool called RAGman that helps manage curriculum “building blocks.” It’s a way for departments to share their specific expertise—such as pedagogical styles or departmental goals—so that it can be combined with general course content.
Real-World Parallels
This isn’t just a theoretical exercise. Other institutions are already testing these waters. The University of Delaware is currently piloting UD Study AiDE, a tool developed by their Academic Technology Services. They are training internal AI models to contextualize course materials using 20 years of proprietary lecture transcripts and LMS data. Their focus is specifically on large foundational courses where students often lose that “personal connection” to the material.
By grounding the AI in their own faculty’s expertise rather than general internet data, they are creating a “verified” layer of personalization that was previously impossible at scale.
Final Thoughts
Duke is a multidisciplinary environment where this kind of cross-departmental synergy is a natural fit. By using the lecture for efficient “general” learning and AI-assisted assignments for “specific” application, we can give students the best of both worlds.
It’s an idea born from a walk with the dogs, but it feels like a viable path toward making general education feel much more personal.
