Large Language Models as communicative actors

• Applicant: Dr. Fabian Anicker
• Subject Area: Sociological Theory
• Term: since 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 553833522

Applied AI research has achieved a significant breakthrough in recent years: the successful implementation of large language models (LLMs). Some of the most well-known examples include ChatGPT, LLaMA, and Gemini. Due to their versatility, they play an outstanding role in the transformation of society through artificial intelligence. At relatively low costs, these systems can adapt content and communication style to the desires of users, thus individualizing mass communication. LLMs are not just the foundation of chatbots or interactive voice assistants. Potentially, they can revolutionize the ways digital technology is used by acting as an interface between people, data, the internet, and other programs. Due to the rapid diffusion of LLM-based applications the social sciences are simultaneously faced with foundational theoretical and empirical challenges. Conventional categories like technology, medium, actor or infrastructure cannot be seamlessly applied to LLMs. Unlike other AI systems and many other technical systems, LLMs can be directly experienced by people in their interactions. To grasp their influence on social processes, the unique mode of their impact must be more precisely determined. What aspects of LLMs can be held responsible for their sometimes strong influence on users? The research project aims to contribute to the experimental investigation of the effects of LLMs on opinions and attitudes and to develop a theory of communicative agency suitable for explaining communication with LLMs.

• DFG Programme: Research Grants
• Cooperation Partner: Professor Dr. Frank Marcinkowski