Visuo-spatial cognition and the redesign of human-AI interaction in generative artificial intelligence

Abstract The rapid integration of generative artificial intelligence (GenAI) into education and professional contexts has produced a growing body of research on human-AI interaction, prompt engineering, and AI-mediated learning. Yet this literature overwhelmingly assumes that all users interact with AI through written text—a fundamentally phonocentric paradigm that excludes deaf learners whose primary language operates in the visuo-spatial modality. This conceptual paper introduces phonocentrism in GenAI as a theoretical lens, arguing that the architecture of Large Language Models, the design of prompting interfaces, and the framing of AI literacy as a text-based competency systematically disadvantage deaf signers. Drawing on deaf studies (specifically the Deaf Gain framework), computational linguistics, and universal design for learning, this paper critically reviews the current landscape of GenAI accessibility research, policy, and institutional practice—revealing the near-total invisibility of deaf learners across all three domains. In response, this paper proposes the visuo-spatial AI interaction model (VSAI), a conceptual framework organized around three dimensions—Input, Processing, and Output—that reconceptualizes human-AI interaction beyond the constraints of linear text. The model is grounded in principles of modality equivalence, cognitive sovereignty, and design from the margins. Implications for teaching practice, institutional policy, AI development, and future research are discussed, with the argument that designing AI interaction for visuo-spatial cognition benefits not only deaf learners but all users whose thinking exceeds the boundaries of phonocentric text.