Friend and foe: Characterising (counter)productive prompting of generative AI to supplement exploratory sketching

Text-to-image Generative AI (TTIGenAI) tools have become commonplace in design practice and education, raising important questions about their role alongside exploratory sketching. This article explores TTIGenAI as an emerging visualisation modality that alters how designers create and explore ideas. Specifically, we examine the acute tensions between GenAI’s near-instantaneous, high-fidelity, text-driven outputs versus the ambiguous, iterative, and reflective act of traditional sketching. We report on a one-day ideation activity by 18 student designers, in which we analysed 1894 prompts, generated imagery, alongside reflective survey data to uncover patterns of productive and counterproductive TTIGenAI use. We outline prompting behaviours that characterise (counter)productive sketching behaviours, including broad exploration, reinterpretation, ambiguity and design fixation. Through our findings, we contribute practical and pedagogical prompting strategies to employ TTIGenAI to support and stimulate traditional sketching. As such, we provide insights on how AI can augment and challenge the cognitive and creative value of sketching.

keywords: generative AI; visualisation; exploratory sketching; prompting strategies; ideation

• with: Charlie Ranscombe (lead), Wendy Zhang, Nabila Afif
• year: November 2025 — ongoing