Abstract
This paper examines the epistemic and ethical dimensions of Luciano Floridi’s (2025a) concept of “distant writing,” a form of AI-assisted composition in which large language models (LLMs) are used to generate, refine, or structure literary and argumentative texts. Drawing on analytic epistemology and virtue theory, and further informed by Floridi's (2025b) thesis on AI as agency without intelligence, it argues that distant writing constitutes a distributed epistemic activity, wherein the human author retains epistemic agency while delegating generative labor to non-agential systems. Central to this analysis is a distinction between epistemic instrumentality and epistemic agency, which clarifies the role of LLMs as sophisticated tools rather than co-authors. The paper explores how procedural, evaluative, and semantic knowledge inform the author’s control over AI-generated content and assesses the epistemic risks introduced by the plausibility and opacity of such outputs. In response to concerns about plagiarism, it proposes a normative reformulation of attribution standards, emphasizing transparency and epistemic conscientiousness over traditional notions of originality. Ultimately, the paper situates distant writing within an emerging epistemic ecology in which creativity, authorship, and intellectual responsibility are collaboratively negotiated between human and machine. It advocates for a taxonomy of AI involvement to guide ethical disclosure and preserve the integrity of authorship in an age of synthetic textual production.