Rethinking Reference and Authorship: On the Philosophical Status of LLM-Generated Verbal Products

Abstract

In this article, the status of LLM-generated verbal products is dis­cussed in principle. While we have so far been socialized in such a way that we automatically assume an intelligent author ‘behind’ verbal products that can be read as intelligent, we can no longer sim­ply assume this close connection in the age of LLMs. In this sense, I name LLM-generated products ‘intelligible textures’. As products, these intelligible textures can hardly, if at all, be distinguished from authorized, human-created texts, but the learning and usage pro­cesses differ fundamentally, especially with respect to acts of refer­ence, particularly acts of exemplification. What consequences does this have for our general conception of the written word, authorship and the ethical-political idea of responsibility for verbal products? This comprehensive question is discussed in the present article us­ing the example of LLM-generated essay evaluations, which seem to be an instructive example here as they require a high degree of judgment and truthfulness.