Frontier Signal
LLM Reasoning Hurts Behavioral Simulation in Multi-Agent Negotiation
New research shows that stronger LLM reasoning can hinder accurate behavioral simulation in multi-agent negotiation, leading to solver-sampler mismatch.
Read the briefing