Investigating the impact of generative AI integration on the sustenance of higher-order thinking skills and understanding of programming logic

This study investigates how integrating generative AI (GenAI) with instructional scaffolding and prompt engineering supports higher-order thinking skills (HOTS) and programming logic. A mixed-methods design was used, combining quantitative and qualitative data. The intervention followed a one-group pretest-post-test structure over seven weeks with 25 computer science students with no prior C++ experience. The GenAI-Ped framework guided the design. It combines Bloom's taxonomy, Seelf-Regulated Learning, Universal Design for Learning, and Vygotsky's Zone of Proximal Development. Students received scaffolded support across six instructional phases, including prompt training and guided GenAI use. Quantitative results showed significant gains in problem-solving (applying constructs: t = 2.38, p = 0.013, d = 0.475), critical thinking (conditional reasoning: t = 2.53, p = 0.018, d = 0.506), creativity (applying new ideas: t = 2.28, p = 0.032, d = 0.456), and programming logic (loops: t = 2.78, p = 0.010, d = 0.555). However, smaller gains were observed in code optimization (t = 1.693, p = 0.103, d = 0.339) and evaluating solutions (t = 1.732, p = 0.096). Qualitative data, including feedback and GenAI chat logs, showed that prompt specificity and scaffolded feedback improved engagement, HOTS, and programming logic. The novelty of the study lies in its demonstration that the integration of GenAI into programming education using GenAI-Ped framework can sustain HOTS and programming logic while mitigating overreliance. These findings offer a practical model for integrating GenAI into programming education.