ai · arxiv/cs.AI · 8 min
LLMs show human-like trust bias toward people, with demographic blind spots
Study of 43,200 experiments reveals language models develop trust patterns similar to humans, including susceptibility to age, religion, and gender bias in financial decisions.
Apr 17, 2026 Read →