Extended Judgment: LLMs and the Architecture of Artificial Capacities

Abstract Large language models (LLMs) are increasingly embedded in everyday epistemic practices, yet their epistemological status remains unclear. Existing debates typically oscillate between treating LLMs as epistemic agents or reducing them to mere tools. This paper argues that both views mischaracterize the core issue. Drawing on a teleological interpretation of 4E cognition and the paradigm of human–AI coevolution, it develops an alternative account that understands LLMs as components of extended epistemic capacities, most centrally the human capacity for judgment. On this view, epistemic capacities are not intrinsic properties of isolated subjects or systems, but purposive and relational achievements enacted within sociotechnical practices. LLMs do not possess epistemic agency or original normativity; rather, they reshape the conditions under which human judgment, inquiry, understanding, and justification unfold. By analyzing language, coevolutionary feedback loops, derived normativity, and epistemic autonomy, the paper shows how epistemic responsibility remains human while epistemic performance becomes technologically extended.