Worldview passport of llm models

non-profit

AI & ML interests

Worldview Passport of LLM Models — a non-profit research initiative studying the worldview layer of large language models. We focus on evaluating how LLMs represent facts, values, norms, and cultural frameworks across societies. Our interests include: Worldview evaluation of LLMs: mapping factual, normative, and cultural components embedded in model outputs. Encyclopedic factual grounding: benchmarking models against structured knowledge bases (BRE, Britannica). Cross-cultural and cross-societal variation: understanding how models shift interpretations across different value systems. Normative transparency: analyzing ethical reasoning without promoting any ideological viewpoint. Humanities-driven AI assessment: integrating methods from philosophy, sociology, anthropology, and cognition studies. Configurable evaluation frameworks: allowing researchers to plug in their own factual corpora and value systems. We aim to bring clarity to the non-technical, human-centered dimensions of AI — the parts that resemble human worldviews.