2026 ASEE Annual Conference & Exposition

The widespread adoption of large language models (LLMs) such as ChatGPT has accelerated
educators’ need to define responsible and effective ways to integrate artificial intelligence
(AI) into authentic learning environments. While early work explored student perceptions
of AI-supported learning, fewer studies have examined structured instructional frameworks
that position AI as a transparent, documented component of student work. Building on prior
research evaluating students’ perceptions of AI’s influence on learning environment, technological
literacy, and teaching effectiveness, the present work implements and evaluates a structured
framework for AI use across multiple assignments in a junior-level software engineering
course.
In this new phase, students were required to engage with LLMs through a transparent and
ethical workflow that constrained the scope of model assistance. Each team-based deliverable
specified: (1) a provided LLM prompt template emphasizing student authorship of ideas and
data; (2) documentation of the complete AI chat history as an appendix; and (3) reflection on
the benefits and limitations of AI’s contributions to the final work. These procedures were
integrated into complex, authentic tasks such as producing software design and database documentation.
The approach aimed to treat AI as a collaborative drafting and reasoning partner
rather than a shortcut to problem completion.
Data were gathered through assignment artifacts, structured reflections, and instructor observations
to examine outcomes across three dimensions: (1) learning environment, as reflected
in team collaboration and engagement; (2) technological literacy, including ethical reasoning,
prompt engineering, and evaluation of AI outputs; and (3) teaching effectiveness, measured by
students’ ability to apply structured AI use while maintaining independent problem-solving.
Preliminary analyses indicate that the structured framework fostered higher accountability and
deeper understanding of AI’s strengths and limitations. Students reported that documenting
their AI interactions clarified the boundary between human and machine contributions and
improved their ability to critically assess LLM outputs.
This work demonstrates how structured integration of AI tools can enhance both student
learning and teaching effectiveness while safeguarding academic integrity. It offers a replicable
model for instructors seeking to move beyond ad hoc AI usage toward systematic, ethically
informed classroom practices. By shifting from perception to practice, this study contributes
to the ongoing dialogue on how AI can responsibly augment—not replace—human cognition
and creativity in engineering education.