Social Connections between Large Language Model Agents : Testing LLM Agents in a Social Setting
- In this study, we explore if large language models (LLMs) can apply the concept of personal closeness and use this to enhance communication between each other while using midrange personal computers. We test and modify the LLM “Llama 3” on commodity hardware to measure the level of interpersonal closeness between the different LLM’s characters. In particular, we show insights into the development of their connection. Results indicate that the scoring of one LLM agent could have an effect on each other's LLM agents’ scoring.
| Author: | Ferdinand Göpfert, Suzan Johannes, Johann Schulenburg, Julian Seibert, Alexander Thier, Thomas Krach, Ruxandra LasowskiGND |
|---|---|
| DOI: | https://doi.org/10.18420/muc2024-mci-src-373 |
| Parent Title (English): | MuC’24, Mensch und Computer, 1.- 4. September 2024, Karlsruhe, Germany - Workshopband |
| Publisher: | Gesellschaft für Informatik e.V. |
| Place of publication: | Bonn |
| Document Type: | Conference Proceeding |
| Language: | English |
| Year of Completion: | 2024 |
| Release Date: | 2024/12/04 |
| Tag: | Interpersonal closeness; Large language model; Social AI; Social simulation |
| Page Number: | 4 |
| Licence (German): |  Urheberrechtlich geschützt |