Can Machines Mean? Wittgenstein vs. Large Language Models

Can Machines Mean? Wittgenstein vs. Large Language Models This paper argues that large language models cannot genuinely mean anything or perform speech acts, lacking the communal forms of life Wittgenstein identifies as necessary for semantic content. Machines inherit and amplify permanent flaws in human language—ambiguity, context-dependence, rule-following paradoxes—without grasping or norm enforcement. Drawing on Philosophical Investigations (§43, §185-242), speech-act theory (Austin/Searle), and LLM case studies, the analysis establishes:1. Fluency ≠ understanding (behavioral mimicry)2. No communal normativity = no meaning3. Policy implication: "AI decided" claims demand heightened liability scrutiny SSRN: 6015815