Scott Franklin has been awarded a Teaching, Learning & Professional Development Center AI Mini Grant
Scott Franklin, associate professor of practice in the Department of Computer Science at Texas Tech University’s Edward E. Whitacre Jr. College of Engineering, has been awarded a Teaching, Learning, & Professional Development Center AI Mini Grant to develop an innovative AI-driven teaching tool aimed at transforming how students learn discrete computational structures.
The project, titled TRACE (Teachable Reasoning Agent for Computational Education), received $2,000 in funding with an additional $1,000 bonus for publication. TRACE uses a GPT-4-powered “virtual student” that flips the traditional AI tutoring script—students don’t get answers from the AI; instead, they teach it.
“Our goal is to be able to scale up the learning-by-teaching strategy to larger classes by having the students tutor the AI instead of the other way around,” Franklin said.
TRACE is specifically designed for CS1382: Discrete Computational Structures and will be piloted in fall 2025 with about 135 students. Unlike standard educational AI tools, TRACE presents common misconceptions and asks questions, prompting students to diagnose errors, explain concepts and deepen their understanding through teaching—an approach backed by research in learning sciences.
The AI’s responses are intentionally naïve, driven by carefully engineered prompts and a knowledge-state tracker that ensures realistic misunderstandings across topics like logic, induction, and probability. As students engage more, TRACE transitions into a more Socratic role, gradually increasing cognitive challenge as student mastery grows.
Franklin’s project aligns with the university’s 2025 strategic initiative, “AI for Assessment & Feedback Innovation,” and will undergo rigorous evaluation in spring 2026. The study will use a randomized controlled trial to compare TRACE users with students completing equivalent assignments without AI assistance. Assessment tools include diagnostic tests, ABET-aligned exam items, metacognitive surveys, engagement metrics and detailed interaction logs.
The research aims to demonstrate improvements in conceptual understanding, self-regulation and student motivation—particularly among lower-performing learners.
Franklin, who brings experience as both an AI developer and computing educator, plans to share results through Texas Tech teaching workshops, a TLPDC “Small Bytes” post, and national computing education platforms, including Special Interest Group on Computer Science Education’s “Nifty Assignments.” The broader goal is to contribute a scalable, evidence-based peer teaching strategy to computer science education.
For more on the grant, visit the Scholarship of Teaching and Learning Mini-Grants section of the Teaching, Learning, & Professional Development Center website.