Investigating the Integration of Generative Artificial Intelligence into Programming Education to Support Higher-Order Thinking Skills and Programming Logic
Year: 2026
Type: dissertation
Abstract: The use of generative artificial intelligence (GenAI) tools in programming education is increasing.Although these tools can support code generation, explanation, debugging, and feedback, unguided use may weaken students' higher-order thinking skills (HOTS), programming logic, and responsible learning practices.This thesis addresses this problem by developing and evaluating the Generative Artificial Intelligence Programming Education (GenAI-Ped) framework as a structured pedagogical approach for integrating GenAI into programming education.The thesis was conducted through three linked studies using design science research and a pragmatic sequential mixed-methods design.Study I systematically reviewed 40 empirical studies on GenAI integration in programming education and used the findings to derive the design requirements for GenAI-Ped.Study II operationalized the GenAI-Ped framework in a seven-week C++ classroom intervention with 25 undergraduate computer science students and designed a HOTS-Programming Logic Rubric, which was used as the primary quantitative instrument for assessing HOTS and understanding of programming logic.Study III evaluated the GenAI-Ped framework through a controlled between-group experiment design with 8 124 undergraduate students learning Java and comparing framework use with unguided GenAI use.Data sources included rubric-scored assessments, learner feedback, and GenAI interaction logs.Study I results showed that GenAI integration in programming education is promising but inconsistent, with recurring gaps in scaffolding, assessment, accessibility, responsible use, and alignment with learning objectives.These findings produced the design requirements for GenAI-Ped.Study II showed improvements in HOTS, such as problem-solving, critical thinking, creativity, and programming logic after the intervention, with qualitative evidence indicating more specific prompting, greater code testing, debugging, and reflection.Study III provided stronger short-term evidence that students who used GenAI through the GenAI-Ped framework achieved higher post-test HOTS outcomes and modestly stronger programming logic outcomes than students who used GenAI without the framework.Process evidence showed stronger pedagogical alignment, cognitive engagement, and responsible-use behavior in the experimental group.Overall, the findings show that GenAI is most educationally valuable when used as a scaffold for reasoning, not as a substitute for student thinking.GenAI-Ped contributes a theoretically grounded and empirically examined framework, with the strongest evidence supporting its scaffolding, promptengineering, verification, reflection, and structured-use components.
Author: Nathaniel, Jemimah
Institution: University of Eastern Finland
Language: English
Subfield: Computer Science Applications
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: green
OpenAlex ID: W7215509865
