Q&A: Exploring generative AI's impact on student thinking during programming and debugging

Dr. Yazid Albadarin's doctoral dissertation titled "Capturing the Impact of Generative AI on Higher-Order Thinking, Creativity, and Cognitive Dynamics During Debugging" will be presented at the Faculty of Science, Forestry and Technology on Joensuu campus. The study investigates how generative AI influences students' higher-order thinking, creativity and cognitive processes during programming and debugging activities.

The research is significant as generative AI becomes increasingly integrated into everyday education. While AI can enhance students' immediate performance, it does not guarantee meaningful and lasting learning. Albadarin's findings reveal that AI-assisted students outperformed and outperformed their non-AI counterparts in some debugging tasks, but exhibited weaker knowledge retention over time.

Moreover, Albadarin's study uncovered that generative AI can reshape students' cognitive approaches. Students who employed AI demonstrated stronger connections between comprehension, application and creativity, while those without AI exhibited stronger ties between analysis and evaluation. Importantly, Albadarin's research underscores that superior performance does not always equate to enduring learning.

These insights are invaluable for educators, researchers and policymakers, emphasizing the necessity of integrating AI into education thoughtfully. The ultimate goal should be to foster students' independent thinking, problem-solving and learning skills, not just to expedite task completion. Albadarin's findings can guide educators in crafting learning activities that promote critical thinking, AI response evaluation and independent problem-solving, rather than merely accepting AI-generated suggestions.

Albadarin's doctoral research comprises three linked studies utilizing both quantitative and qualitative methods. The first study involves a systematic literature review of 14 empirical studies on ChatGPT in education, identifying key research gaps. The second study employs a quasi-experimental design with 152 university students, comparing those who used generative AI (LLAMA 3.1) during programming and debugging tasks with those who worked without AI.

Data were gathered through assessments, surveys, interviews and a retention test conducted seven weeks later. The third study involved 132 students, employing cognitive network analysis to examine the relationships among comprehension, application, analysis, evaluation and creativity. Collectively, these studies illuminate how generative AI impacts students' performance, thinking processes, creativity and long-term learning.