In This Story
AI tools can generate code, explain existing programs, and assist with software development. Since beginner programmers could become overly dependent on AI, educators are now tasked with helping students use it effectively without sacrificing the fundamental skills they need to become successful computing professionals.
Marcos Zampieri, assistant professor in the Department of Information Sciences and Technology in the Long Nguyen and Kimmy Duong School of Computing at George Mason University, is a principal investigator in two new National Science Foundation–funded projects designed to address that challenge.
Together, the projects focus on helping students learn to work with AI responsibly, securely, and in ways that support rather than replace learning.
"If students use AI for absolutely everything and not in a pedagogically sound way, they will not acquire the foundational skills that computing professionals need to succeed," said Zampieri.
Building a FRAMEwork
The first project, funded through NSF’s Innovation in Two-Year College STEM Education (ITYC) program, brings together researchers from George Mason, Rochester Institute of Technology, and Northern Virginia Community College (NOVA). The team will develop FRAME, a framework for integrating large language models (LLMs) into introductory computer science courses. The project will create instructional materials for NOVA’s Introduction to Problem Solving and Programming and Object-Oriented Programming courses.
“We are proposing here to develop a framework to integrate large language models into computer science curricula,” Zampieri said. “The students learn in a pedagogically aligned way to work with LLMs for software development.”
Ask ERICA
A second NSF-funded project takes that same philosophy into cybersecurity education. Known as ERICA, the Educational Security-Focused Intelligent Coding Agent project brings together researchers from George Mason, University of Notre Dame, and The University of Texas at Arlington to develop an AI-powered assistant that promotes secure coding practices.
The project will investigate how trustworthy AI coding assistants can support undergraduate software security education while incorporating safeguards that prevent the generation of harmful or malicious code. “We’re framing this as investigating how trustworthy AI coding agents can support undergraduate secure coding education,” Zampieri explained.
Both projects will make their frameworks, educational materials, and software openly available to educators. “The framework, the assignments, anything we create will be open. Also, the software that we create as part of the projects will be open,” said Zampieri.
With research expertise in natural language processing and LLMs, Zampieri aims to help educators understand how AI can support learning while ensuring students still develop core computing skills.
Through FRAME and ERICA, Zampieri and his collaborators hope to create models for responsible, pedagogically aligned AI use that can benefit computing programs well beyond George Mason and its partner institutions.