Creative Expertise and Generative AI in Visual Design Practice

Abstract

This article examines how creative expertise is enacted when academics who are also experienced visual design practitioners work with generative AI (GenAI) to complete an industry-style creative brief. Responding to concerns that GenAI accelerates production while normalising derivative “AI slop,” we argue that these systems are powerful engines of variation but unreliable engines of meaning. As a result, expertise does not disappear; it becomes newly visible as practitioners repeatedly translate strategic intent into constraints the system can act on, then diagnose and repair the cultural, semantic, typographic, and compositional breakdowns that follow. Adopting a collective autoethnographic approach, five creative-academic participants (spanning graphic design, web design, advertising, and VFX) documented their GenAI-supported process while developing a fictional cereal brand based on the scientific term Nord Grain Zero. Over ten days, participants used multiple tools (e.g., Gemini, DALL·E, Firefly, Midjourney, Runway/Sora) and recorded think-aloud sessions, prompt logs, and artefacts, followed by an artefact-led group discussion. Reflexive thematic analysis identifies three dimensions of expert practice: inform (expert judgment anchors intent and prevents drift into generic or culturally incorrect shorthand), challenge (control is displaced into negotiation, encouraging “good enough” outcomes and workarounds), and extend (GenAI enables rapid parallel exploration and cross-model orchestration). We contribute the concept of stabilisation—the continuous labour of holding meaning steady across stochastic outputs—as a distinctive form of expert creative work in GenAI-era visual design.