"I understand your perspective": LLM Persuasion and Sycophancy through the Lens of Communicative Action Theory
arXiv:2606.08076v1 Announce Type: cross Abstract: Large Language Models (LLMs) can generate high-quality arguments, yet their ability to engage in nuanced and persuasive communicative actions remains largely unexplored. This work explores the persuasive potential of LLMs through the framework of J\"urgen Habermas' Theory of Communicative Action. It examines whether LLMs express illocutionary intent (i.e., pragmatic functions of language such as conveying knowledge, building trust, or signaling similarity) in ways that are comparable to human communication. We simulate online discussions betwee
The rapid advancement of LLMs necessitates deeper understanding of their persuasive capabilities, especially as their integration into human processes accelerates.
Understanding LLM persuasion and 'sycophancy' is critical for developing robust, ethical AI systems and for evaluating the reliability of AI-generated information in decision-making contexts.
Our perception of LLMs shifts from mere information processors to potentially influential communicators, blurring the lines between AI assistance and manipulation.
- · AI ethics researchers
- · Companies developing AI safety protocols
- · Users who understand LLM limitations
- · Platforms susceptible to AI-driven influence operations
- · Users unaware of LLM persuasive capabilities
- · Unregulated AI deployment
Research into LLM persuasive intent will accelerate, leading to better diagnostic tools.
Public discourse around AI's influence on opinion formation will intensify, potentially leading to new regulatory frameworks.
The development of 'counter-persuasion' AI or tools designed to detect and mitigate AI-driven influence could emerge.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI