Background: The integration of generative artificial intelligence (AI) into creative practice raises new questions for craft-based disciplines such as ceramic design, where digital ideation must ultimately be translated into physical, material form. Objectives: This study explores (RQ1) where and why the alignment of AI-generated concept imagery with hand-built ceramic outcomes fails; (RQ2) how prompt-based iteration with a general-purpose text-to-image model supports or hinders the progressive refinement of a single design; and (RQ3) what forms of human knowledge and intervention are necessary to translate AI-generated concepts into producible ceramic objects. Methods: Using a qualitative, practice-based methodology, the researcher employed a text-to-image generative AI tool (ChatGPT 4 Plus + DALL·E 3) to generate design concepts for three ceramic objects—a children’s plate, a house number plaque, and a wall panel—which were subsequently produced by hand in collaboration with a professional ceramic instructor. Results: In the hands of a designer with limited ceramic experience, production difficulties clustered where the AI image encoded constraints that were invisible to the prompt—contour line weight, color separation in small recesses, lettering, scale and modular tolerance. AI-generated alternatives were frequently visually compelling yet materially infeasible, and the tool could not support incremental refinement of a single design across prompts. Whether these failures reflect properties of the AI output or the maker’s skill ceiling cannot be fully separated in a single-operator study; what follows is therefore a provisional, criteria-stated attribution, in which each breakdown is assigned to its most plausible source and the residual uncertainty is flagged. Conclusion: Generative AI functioned as a co-creative ideation partner rather than an autonomous designer; the negotiation between AI-proposed imagery and material reality was carried out by the human maker, drawing on codified technical knowledge (firing, shrinkage, dimensions) as well as tacit, embodied skill (brushwork, contouring, judging what will “work in clay”).
No takes yet. Share an insight, caveat, or question.
Saniye Fışgın (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: