Metasystem Transitions in Education: Human-AI Interactions as Catalysts for Computational Thinking - Figure 1

The analysis of post-intervention data revealed that scaffolded use of generative AI in project-based learning was associated with significant improvements in students’ CT. Specifically, students in the experimental group, who were guided to take an agentive role in their interactions with Gen-AI tools, demonstrated a greater increase in CT scores compared to their peers in the control group. The average CT score in the experimental group rose from 34.3% to 41.2% over the academic year, representing a 6.9 percentage point gain. In contrast, the control group, which used Gen-AI tools without targeted scaffolding, showed a smaller increase from 30.4% to 33.5%, a 3.1-point improvement. (see Figure 1).