Subjektsimulation und Diskursethik

This essay asks whether contemporary conversational AI can count as a parti­cipant in discourse-ethical communication. Starting from Searle’s Chinese Room and the “semantic turn” brought about by transformer-based Large Language Models, I argue that machine-generated utterances now provide convincing functio­nal equivalents of semantic understanding. To assess their status in com­mu­nication, I map human–AI interaction onto Schulz von Thun’s four-sides model and then test it against Habermas’s three validity claims (truth, sincerity, rightness). AI systems perform strongly on truth-oriented tasks but lack sincerity in the strict sense; a defensible proxy is transparency about limits, provenance, and uncertainty. Rightness depends on human curation and governance rather than machine-held convictions, raising concerns about operator power and censorship. In the final part, I reinterpret public AI as a technological re-objectification of shared social meanings: a “personalized avatar” of Mead’s Generalized Other rooted in Hegel’s objective spirit. On this view, AI is not a subject but a new medium of collective self-observation, whose legitimate role in discourse hinges on public accessibility, contestability, and plural control. The upshot is a shift from questions of agency to questions of institutional design and normative inte­gration. Keywords: artificial intelligence; discourse ethics; Habermas; generalized other; objective spirit; Mead; Hegel; large language models; semantic understanding; public sphere.