Disembodied creativity in generative AI: prima facie challenges and limitations of prompting in creative practice

Abstract

This paper examines some prima facie challenges of using natural language prompting in Generative AI (GenAI) for creative practices in design and the arts. While GenAI is purported to “democratize” creativity by offering a new mode of creation, we argue that it comes with a significant mortgage—particularly one in relation to expert performance, skill acquisition, and embodied engagement. Drawing from Dreyfus and Dreyfus, we show that creativity grounded in internalized expert knowledge cannot be reduced to rule-following or meaningfully externalized in instructions, i.e., prompts. Building on Polanyi, Simon, and Sennett, we posit that much of what makes creative work meaningful is tacit and intuitive, and therefore cannot be fully articulated through prompts. From the perspective of embodied and enactive cognition (Thompson, Noë, Pallasmaa), we argue that even “traditional” digital tools retain a material, bodily interface—something entirely absent from prompt-centered creation. While it may be tempting to treat GenAI systems as mere instruments, the mode of interaction they afford introduces a discontinuity: unlike analog tools or conventional software, they offer the creator significantly less control and disrupt and even erodes the feedback loop between mind, hand, and expressive material. Rather than supporting skill development, prompting risks sequestering the user in novice-level engagement. By addressing these challenges, our analysis offers a clearer view of what is at stake when generative systems are integrated into creative disciplines, and why human creators, integrating multiple creative and epistemic faculties as they see fit, must remain at the center of that process.