Evaluating AI-assisted creative ideation: A crossover study in higher education

As artificial intelligence (AI) becomes increasingly integrated into educational contexts, its impact on students’ creative thinking remains unclear. Given the critical role of creativity in engineering and design education, understanding how AI tools shape students’ ideation processes is essential for developing effective pedagogical practices. This study examines the impact of generative AI, specifically ChatGPT, on creative ideation among undergraduate design engineering students. The research was conducted through a randomised crossover experiment, with students alternating between AI-assisted and unaided ideation tasks. A mixed-methods approach combined quantitative analyses of the 728 ideas generated with a qualitative evaluation of students’ interactions with AI. Results show that the use of AI did not reduce fluency, flexibility, or originality, nor did it lead to thematic homogenisation. However, semantic divergence was significantly lower in the AI-assisted condition, suggesting convergence in the way ideas were formulated. Additionally, AI-assisted ideas more frequently resembled existing products and exhibited reduced textual elaboration. A mixed between-within subjects ANOVA revealed that students who began with AI support produced more original and diverse ideas across both tasks, pointing to a lasting effect of early AI use. Qualitative analysis of student-AI interactions revealed important patterns, including predominantly passive and directive use, with limited exploratory or collaborative engagement. These findings provide new insights into human-AI co-creation and highlight the importance of promoting intentional, critical, and pedagogically guided use of generative AI tools in education.