Domesticating robots, domesticating language: The hylomorphic paradox of “openness” in large language models

This article examines how openness is interpreted in human–LLM interaction through an ethnographic study of a robotics experiment. Focusing on an episode in which a robot produced an unexpected utterance, I analyze how engineers classify the output as inconsequential noise. I argue that such dismissal reflects a hylomorphic understanding of language, where form is externally imposed on linguistic matter, by latching onto ideas of alignment and usefulness. I configure such hylomorphic movement as undergirding the semiotic ideology at play in robotics engineering, one which constrains the dynamism and potential of situated interaction, transforming openness into a domesticated variability, and limiting the emergence of forms of coordination and meaning.