1 paper · 1 filter
Yu Sha, Junqi Tao, Dixin Zhou +10
Large language models (LLMs) increasingly mediate human decisions and communication, yet their behavioural regularities remain difficult to characterize systematically. We develop…