The role of anthropomorphism in technical discourse: linguistic objectification of machines and algorithms

Abstract

This article examines the role of anthropomorphic language in contemporary technical discourse, with particular focus on texts describing artificial intelligence systems and computational algorithms. Anthropomorphism – the attribution of human traits, intentions, emotions, and cognitive capacities to non-human entities – has become a pervasive structuring feature of the language used to describe AI across academic publishing, corporate communication, and journalistic reporting. Rather than treating this phenomenon as a peripheral stylistic matter, the study approaches it as a substantive epistemic, rhetorical, and ethical issue with far-reaching consequences for how intelligent technologies are understood, trusted, regulated, and governed. The article proposes a three-part linguistic taxonomy of anthropomorphic expression in technical registers, distinguishing between lexical anthropomorphism – the deployment of mentalistic vocabulary such as knows, decides, understands, and feels – syntactic anthropomorphism, by technical systems are positioned as agentive grammatical subjects of volitional predicates, and metaphorical framing, through which sustained conceptual mappings such as computer as mind organise entire disciplinary vocabularies. Drawing on naturally occurring examples AI research literature, technology company communications, and mainstream journalism, the analysis demonstrates that anthropomorphic language is not incidental to these discourses but is institutionally reproduced and commercially incentivised within them. The study further identifies the principal functional drivers of anthropomorphic usage: communicative economy, conceptual scaffolding for non-specialist audiences, commercial persuasion, and disciplinary inertia. It argues, however, that these pragmatic advantages carry significant epistemological and ethical costs. Most critically, anthropomorphic framing systematically misrepresents the locus of agency in sociotechnical systems, displacing accountability the human designers, deployers, and regulators of AI onto the systems themselves. It also distorts public risk perception in ways that simultaneously foster unwarranted trust and activate culturally embedded fears of malevolent machine agency – both of which impede informed public deliberation. The article concludes by calling for the development of critical metalinguistic awareness as a professional and civic competency, and identifies corpus-based, experimental, and cross-linguistic research as priority directions for future scholarly investigation.