A World Values Survey-grounded simulation of LLM agent societies found that more than 50% of personas fail to express their assigned WVS value profiles from the outset, while only 2-7% drift after rep…
Researchers propose ExpertIVS, a framework using 14 Sociological Expert Agents to interpret World Values Survey responses, achieving 90.78% value restoration fidelity and outperforming baselines by 5.…
A new arXiv paper (arXiv:2608.09937v1) applies cultural consensus theory (CCT) to analyze how large language models (LLMs) align with cultural norms, finding that models often misrepresent cultural st…
Chinese researchers from the University of Science and Technology of China, Tsinghua University, and Fudan University have developed Light Society, a framework that simulates societies of over one bil…
A new benchmark study from arXiv preprint 2607.26348 finds that large language models (LLMs) fail to replicate real human survey responses across two independent domains—U.S. general social attitudes …
A new study testing GPT-5.4, Claude Sonnet 4.6, Gemini 2.5 Flash, and Qwen3-235B on 101 World Values Survey questions across 13 language-country slices found that prompt framing is a first-order deter…
A new study finds that prompt robustness in large language models depends on the type of question asked, with objective questions showing different sensitivity to prompt variations than subjective bel…
Researchers have developed a framework for evaluating and adjusting the cultural values of large language models by analyzing their responses to scenario-based dilemmas rather than direct questioning.…