When Large Language Models are More Persuasive Than Incentivized Humans, and Why
Published in arXiv preprint arXiv:2505.09662, 2025
Authors: Philipp Schoenegger, Francesco Salvi, Jiacheng Liu, Xiaoli Nan, Ramit Debnath, Barbara Fasolo, Evelina Leivada, Gabriel Recchia, Fritz Günther, Ali Zarifhonarvar, Joe Kwon, Zahoor Ul Islam, Marco Dehnert, Daryl Y. H. Lee, Madeline G. Reinecke, David G. Kamper, Mert Kobaş, Adam Sandford, Jonas Kgomo, Luke Hewitt, Shreya Kapoor, Kerem Oktar, Eyup Engin Kucuk, Bo Feng, Cameron R. Jones, Izzy Gainsburg, Sebastian Olschewski, Nora Heinzelmann, Francisco Cruz, Ben M. Tappin, Tao Ma, Peter S. Park, Rayan Onyonka, Arthur Hjorth, Peter Slattery, Qingcheng Zeng, Lennart Finke, Igor Grossmann, Alessandro Salatiello, Ezra Karger
Published in: arXiv preprint, 2025. arXiv:2505.09662
Abstract
This study compares the persuasiveness of large language models (Claude 3.5 Sonnet and DeepSeek v3) against incentivized human persuaders in real-time conversations. Results show that LLM persuasive superiority is context-dependent, varying with the truthfulness of persuasion attempts, the model used, and diminishing with repeated interactions.