Reimagining service delivery in the agentic era with Google Public Sector
cloud.google.com·21h ago
AI Research
arXiv CS.AI·Jun 9·5 min read
"I understand your perspective": LLM Persuasion and Sycophancy through the Lens of Communicative Action Theory
TL;DR
Large Language Models (LLMs) have shown potential in generating arguments, but their persuasive capabilities are not well understood. This research applies Jürgen Habermas' Theory of Communicative Action to analyze LLMs' illocutionary intent, which refers to the intended meaning behind statements.
✦ Why It Matters
Engineers can leverage insights on LLM persuasion to enhance user interactions in applications requiring effective communication.
Key Takeaways
- Explore LLMs for generating persuasive content in discussions.
- Utilize LLMs to craft arguments that align with user perspectives.
- Train LLMs on human preferences to improve communication effectiveness.
- Evaluate LLM-generated content for higher user agreement.
How It Works
The study employs Jürgen Habermas' Theory of Communicative Action to analyze how LLMs express illocutionary intent, which refers to the intended meaning behind statements. By simulating discussions, the researchers assess how well LLMs can mimic human-like persuasive strategies, particularly in crafting arguments that resonate with users' beliefs.