The rapid assimilation of generative AI technologies into computer science education has prompted growing interest in how institutions adopt and regulate these tools, how students engage with them across varied learning contexts, and how adaptive tutoring systems can be innovated to enhance educational outcomes. This thesis presents three interconnected studies on generative AI in computing education. The first study provides a longitudinal syllabus analysis that explores the assimilation of generative AI policies from Spring 2024 to Fall 2025 for a sample of 148 computer science course syllabi from Hood College by employing a hybrid approach that combines rule-based inference and embedding-based semantic similarity (all-MiniLM-L6-v2) with a classification accuracy of 90. 56%. The key findings of the study show that policy assimilation increased from 12. 8% to 58. 8% during the four semesters, reflecting a substantial institutional shift in how generative AI is formally addressed in computing curricula. The second study provides an analysis of student self-reflections from "Programming Languages: Their Design and Compilation" by employing a semantic NLP pipeline that filters AI-relevant sentences and applies sentence-transformer matching (all-MiniLM-L6-v2) with supplementary VADER sentiment and keyword signal metadata to categorize AI usage patterns into three different trajectories: COLLABORATIVE, OVERRELIANT, and RESISTANT reliance on AI technologies. The third study provides an exploration of the EduAI tutoring system that employs a hybrid approach that combines the Flask web development tool with the LLaMA 3. 3-70B language model via the Groq API and a TF-IDF retrieval-augmented generation (RAG) engine with six levels of adaptive scaffolding and seven behavioral scenarios to support students in developing independent learning habits and reducing over-reliance on AI. Together, these studies advance understanding of how generative AI is reshaping teaching practices, student learning, and educational system design in computing education. Keywords: generative AI, over-reliance detection, adaptive tutoring, NLP, computing education, scaffolding, syllabus analysis
Ngoc Bui (Wed,) studied this question.