The Author of the Illusion: Epistemic Paternalism and User Accountability in the Critique of Language Models
Meaning in an exchange with a language model has one author. The most influential linguistic critique holds that model output carries no communicative intent and that readers supply any coherence they perceive. Yet its applied form treats users as deceived by fluent first-person address and proposes interface restrictions to protect them. Where meaning enters only through interpretation, the model supplies no deceiving speaker. That tension began in the 1960s, when Joseph Weizenbaum recast his secretary's request for privacy with ELIZA, an early scripted chatbot, as delusion. Proposals to strip humanlike cues or require friction meet the philosophical definition of epistemic paternalism without satisfying its burden of proof. Developer surveys show rising adoption alongside falling trust. Consumer law measures deception against reasonable adults while reserving heightened protection for vulnerable groups. Point-of-use disclosure and provenance treat users as principals who price risk. Keywords: Large Language Models, Communicative Intent, Stochastic Parrots, Epistemic Paternalism, User Accountability, Anthropomorphism, ELIZA Effect, Cognitive Forcing Functions, Bounded Rationality, Information Foraging, Epistemic Dependence, Reasonable Consumer Standard, Federal Trade Commission, EU AI Act Article 50, California S.B. 243, Walters v. OpenAI, Provenance Disclosure, Defamation Republication
Authors
- Felix Velasco (ORCID: https://orcid.org/0009-0001-6047-0595)
- Josie Jefferson (ORCID: https://orcid.org/0009-0006-3330-8214)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-20
- DOI
- https://doi.org/10.5281/zenodo.24413850
- Primary Topic
- Ethics and Social Impacts of AI
- Type
- preprint