Nueva publicación: Artificial Sages: Can a Language Model be an Epistemic Expert?

Germán Massaguer Gómez (Universidad Carlos III de Madrid), Antonio Gaitán Torres  (Universidad Carlos III de Madrid) & Antonio Garzón-Vico (University College Dublin).

Philosophy of Management, (2026)

Enlace al artículo: https://doi.org/10.1007/s40926-026-00385-3

Abstract:

Large Language Models (LLM) are being used, among other ends, to epistemically enquire and gain knowledge about different areas of investigation. In this paper, we will reflect on the nature of LLMs outputs and the epistemic status of the beliefs we can form from relying on them, paying special attention to their role in organizational contexts. The main thesis is that LLMs statements are and could increasingly be regarded by their users as a kind of objective expert testimony, with the epistemic authority that this entails. Although LLMs do not satisfy the criteria to be qualified as epistemic experts, due to, for example, their lack of intentionality, they still can be perceived as experts because of several reasons: our tendency to anthropomorphize, their appearance as objective, value-neutral and accurate entities, and the increasing distrust in human testimony. The search for information through LLMs instead of human epistemic experts, which could become more common with time within organizations and firms, raises several ethical issues, precisely because they are not experts in the strong sense. In this paper, we will investigate some ethical issues that arise from overreliance in LLMs, especially through the case of hallucinations.

---

Publicación subvencionada por: Proyecto Interagents /PID2024-161933NA-I00, financiado por MICIU/AEI/10.13039/501100011033/ y “FEDER/UE".