Embodiment as Mirage: The Limits of Artificial Intelligence in Contrast to Human Cognition

Abstract This article examines the epistemological and ontological distinctions between human intelligent behaviour and the technical embodiment found in contemporary artificial intelligence. Using the scene in Ex Machina where Nathan and Caleb discuss Ava’s cognition before Pollock’s No. 5, 1948, we propose an epistemological allegory: while Pollock’s action painting exemplifies relational, situated, and emergent intelligence grounded in embodied coupling with the world, Ava’s intelligence remains anchored in symbolic architectures and computational representations. Drawing on Maturana and Varela’s biology of cognition, enactivist accounts of sensorimotor coupling, Craver’s multiscale mechanisms, and recent advances in teleoperation and social robotics, we argue that embodied AI continues to operate within an individualist ontology that conflates technical embodiment with lived embodiment. This conflation constitutes what we call a mirage of embodiment: the projection onto machines of attributes that, in humans, arise from historical, affective, and intercorporeal couplings. We show that human intelligent behaviour emerges not from internal computation but from dynamic networks of structural and intercorporeal relations that sustain meaning-making. The article concludes by proposing a shift toward relational ontologies in AI design, an approach based on sociotechnical assemblages and architectural foresight, to avoid reductionist interpretations of embodiment and to guide the development of artificial systems capable of participating, rather than simulating, meaningful interaction.