Behavioral Inference at Scale: The Fundamental Asymmetry Between Motivations and Belief Systems. LLM agents reveal funda...

Behavioral Inference at Scale: The Fundamental Asymmetry Between Motivations and Belief Systems. LLM agents reveal fundamental asymmetry in inferring motivations vs. belief systems from behavior, with transformers hitting 34% accuracy ceiling in belief system extraction-implications for AI alignment and safety.Source: arXiv cs.MAhttps://arxiv.org/abs/2509.05624#MachineLearning