GenAI-Assisted Learning in Online Problem-Solving Assessments: Cognitive and Behavioral Insights Across Course Levels

Generative Artificial Intelligence (GenAI) agents play an increasing role in education, particularly in complex problem-solving and analytical programming. This study explores GenAI usage across different levels of Computer-Based Problem-Solving Assessments, revealing that analysis-focused queries dominated initially, with application-level queries prevailing. As task difficulty increased, GenAI reliance grew, with more application-level queries but fewer knowledge-level ones. Effective strategies such as step-by-step decomposition and precise questioning with the Chain-of-Thought approach can enhance GenAI-assisted learning, improving problem-solving efficiency.