Abstract
Artificial intelligence (AI) tools are extensively utilised in design education. Despite the current intelligent-TPACK framework's ability to assess educators' knowledge integration, a deficiency remains in the empirical understanding of the interaction mode. This study integrates log analysis and semi-structured interview data to develop a user-centred mixed-methods approach, aiming to provide empirical evidence and theoretical foundations for future AI in design education through the Intelligent-TPACK framework. The results demonstrate that In terms of teacher operations, educators primarily utilise AI tools for optimising design tasks and producing knowledge visualisations; Students spent an average of approximately 55 minutes using AI tools to complete Design Task 1, with commonly employed tools including Doubao(豆包), ChatGPT, and Midjourney; for Design Task 2, the average AI tool usage was approximately 54 minutes, with Midjourney and DeepSeek being frequently utilised. Within the classroom interaction dimension, the primary behavioural pattern observed was teacher–student co-creation with AI, involving collaborative exploration of design innovation. In the classroom interaction dimension, the primary focus was on the teacher and students collaboratively using AI tools to generate and refine visual concepts for a project on intangible cultural heritage, discussing design choices and iteratively improving the output. The research data empirically validates the HCI-driven design education research theory and the TPACK theoretical framework, establishing a replicable research paradigm with significant reference value for design education research.