Large Language Models Reflect the Ideology of their Creators

Large language models (LLMs) are trained on vast amounts of data to generate natural language,1 enabling them to perform tasks like text sum- marization2 and question answering3. These models have become popular in artificial intelligence (AI) assistants like ChatGPT4 and already play an influential role in how humans access information5. However, the behavior of LLMs varies depending on their design, training, and use.6 In this paper, we uncover notable diversity in the ideological stance exhibited across different LLMs and languages in which they are accessed. We do this by prompting a diverse panel of popular LLMs to describe a large number of prominent and controversial personalities from recent world history, both in English and in Chinese. By identifying and ana- lyzing moral assessments reflected in the generated descriptions, we find consistent normative differences between how the same LLM responds in Chinese compared to English. Similarly, we identify normative disagree- ments between Western and non-Western LLMs about prominent actors in geopolitical conflicts. Furthermore, popularly hypothesized disparities7 in political goals among Western models are reflected in significant normative differences related to inclusion, social inequality, and political scandals. Our results show that the ideological stance of an LLM often reflects the worldview of its creators. This raises important concerns around tech- nological 8 and regulatory 9 efforts with the stated aim of making LLMs ideologically 'unbiased', and it poses risks for political instrumentalization.