Reproduce “Beyond Epistemia: Epistemic Schizologia and Large Language Models as Techno-Semiotic Machines” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Federico Cabitza, Gianluca Colombo
Researchers introduce "epistemic schizologia" to conceptualize the cleavage between LLM-generated plausible language and its warranting semiotic practices, arguing against internalist views of AI epistemology. The work proposes understanding Large Language Models as "techno-semiotic machines" and outlines a practice-centered framework for designing responsible human-AI interactions.
Key insights
- LLMs are best understood as "techno-semiotic machines" that statistically reactivate structures of human writing, generating fluent text without possessing internal understanding or responsibility.
- "Epistemic schizologia" explains how epistemically recognizable discourse can circulate detached from its "warranting semiosis," which refers to the practical genealogy of observations, evidence, and accountable acts.
- The illusion of knowledge (Epistemia) is perpetuated by "eikotic closure" (presenting plausible output as final) and "algorithmic authority" leading to "epistemic self-misrecognition" by users who equate persuasive formulation with justified knowledge.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.