Classroom Reflections on AI-Assisted Creativity in Design Education

Abstract

Debates around Artificial Intelligence (AI) and creativity often split between promise and risk. In this reflective paper, we take a practice-first view from two implementations in design/art courses at a Malaysian public university. We focus on two themes that run through public and official discourse: the need for governance/ethics; and for treating AI as a tool, not the source, of creativity. In the first implementation, the introduction of AI lacked clear guardrails, which led to confusion about acceptable prompts, uneven quality, and uncertainty around attribution. In the second, we introduced explicit instructions (when to use, how to use, and what to avoid), process evidence requirements, and assessment that foregrounded decision-making. Across both, we observed consistent benefits aligned with studio needs: faster divergent idea generation, access to large reference sets, rapid prototyping, and more engaging critique. Risks centred on over-reliance and aesthetic homogenisation, which were mitigated through iteration limits, disclosure of what was student-led versus AI-assisted, and targeted AI-literacy supports. The implementations drew on three categories of tools: a large-language-model assistant for brainstorming and critique, text-to-image generators for fast ideation, and assistive image tools for workflow tasks. We distil these lessons into a seven-part framework for implementing AI in design courses (positioning, guardrails, process evidence, assessment, AI literacy, equity supports, governance/ethics). We conclude that AI can enhance student creativity when governance is explicit and authorship remains central. Limitations include a single institution and a short time frame; future work should test the framework across cohorts and newer models.