Counterparty Modeling is Not Strategy: The Limits of LLM Negotiators
Why advanced reasoning in LLMs does not equate to strategic competence in multi-turn negotiations.
Negotiation requires more than inferring what the other side wants: it requires using that information to make advantageous offers and counteroffers over multiple turns. We study whether large language model (LLM) agents do this in a controlled multi-attribute bargaining environm...
ArXiv CS.AI
Source
Generated insight by OnAirToday AI
Current LLM agents excel at identifying counterparty preferences but lack the sophisticated strategic planning required to translate that information into superior bargaining outcomes. This research highlights a critical gap where linguistic fluency fails to bridge the transition from social modeling to genuine competitive decision-making.
Comments
Keep comments relevant and respectful. No links, advertising, self-promotion, or spam.