References

Since early 2025, a prototype authorial language model based on the culturally sedimented position of Johann Wolfgang von Goethe has existed under the name “GoetheGPT.” What began as a project out of interest in large language models and the innovative use of diverse data sets became the subject of this thesis in the “Semiotics and Multimodal Communication” program at Chemnitz University of Technology. Against the backdrop of scientific debate on issues of digital authorship, artificial consciousness, and the production of meaning by large language models, it has opened up new perspectives on the functionality and scope of the application. After numerous stages of development and insights, a usable interim result has now been achieved that not only generates texts in the style of the poet and thinker, but can also produce new meanings as an interpretable instance in the mode of machine semiosis. As an operational middle ground between historical referentiality and algorithmic generalization, the application moves between philological practice and machine-generated meaning, thereby shifting the concept of authorship from its romantic origins to a rule-based position in the process of meaning formation.