PhD Dissertation Proposal: Boming Zhang, Reimagining Computer Science Education in the Age of Large Language Models
Content
Speaker:
Abstract:
The rise of large language models (LLMs) such as ChatGPT, Claude, and Cursor is rapidly reshaping the landscape of computer science (CS) education. These tools are increasingly used by students for programming assignments, debugging, brainstorming, and project development. However, their widespread adoption has also raised urgent concerns around academic integrity, student learning outcomes, and the future of instruction. Studies have shown a marked shift from traditional forms of plagiarism to the use of generative AI, with many educators struggling to detect or address LLM-based cheating. In this new landscape, simply banning or avoiding LLMs is neither practical nor sustainable.
This dissertation addresses a central question: How should computer science instructors redesign course structures and pedagogy in response to the widespread availability of LLMs? Instead of resisting the presence of these tools, this work explores how they can be used productively, ethically, and equitably to enhance student learning.
The project will examine the following sub-questions:
RQ1: Curriculum Redesign: How should early computer science curricula evolve in an age of agent-based AI assistance? What new learning objectives, aligned with frameworks like Bloom’s taxonomy, are needed to foster critical thinking and human-AI collaboration?
RQ2: Responsible Use: How can we promote fairness, transparency, and responsible use of LLMs in ways that support all learners?
RQ3: Learning Gains: What strategies for incorporating LLMs into the classroom lead to measurable improvements in student learning outcomes?
RQ4: Project-Based Learning: Can LLMs support students in undertaking more complex, authentic, and open-ended projects that would otherwise be too difficult to manage?
This research includes the design and evaluation of novel pre/post assessments and learning interventions across both introductory and graduate-level CS courses. Outputs will include academic publications as well as an open-access repository of practical, vetted LLM-integrated teaching strategies. Through this work, the dissertation aims to provide actionable frameworks and empirical evidence to help educators harness LLMs not as threats, but as transformative tools for the future of computer science education.
Advisor:
Ivon Arroyo