Xian Wu

RAG models struggle to use retrieved knowledge even when it's relevant, but GuarantRAG's two-stage generation and joint decoding boosts accuracy by 12% and slashes hallucinations by 16%.

LLMs exhibit a "Utopian bias" when simulating human behavior, converging towards an unrealistic "positive average person" and failing to capture individual differences and long-tail behaviors.

Today's best AI agents can only complete 33% of common online tasks like booking appointments or filling out job applications, revealing a significant gap between current capabilities and real-world utility.

LLMs can cut code editing costs by up to 50% simply by knowing when *not* to guess.