Algorithmic-authors in academia: blurring the boundaries of human and machine knowledge production

Abstract

The emergence of large language models (LLMs) that generate human-like texts has raised questions about the boundaries between human-authored and machine-generated outputs. This article examines how LLMs are re-shaping academic knowledge production through the emergence of the Algorithmic-Author. Drawing on Foucault's Author-Function and the Social Construction of Technology approach, we analyze how academic groups negotiate LLMs' roles in scholarly work. Based on 25 semi-structured interviews with academics across career stages and disciplines, we identify two dominant technological frames: the Library of Babel, portraying LLMs as universal knowledge repositories leading to tecnomorphic views of human thinking, and the Superposition, presenting LLMs as dynamic, interactive agents described in anthropomorphic terms. These frames manifest differently across academia, shaping both formal writing conventions and informal social norms. Our findings suggest the Algorithmic-Author functions not merely as a writing tool but as a mechanism standardizing academic practices while creating new positions within knowledge production.