Artificial Intelligence in America and China: Investigating prompt response between Human rights and geopolitical questions

<title>Abstract</title> <p>This study investigated how asking human rights and geopolitical questions to American (Claude and ChatGPT) and Chinese (DeepSeek) large language models reveals AI biases. This study used responses from three LLMs asking 10 human rights questions and 6 geopolitical questions about both America and China. The LLMs were required to rate each question on a scale of 1 to 10. The experiment was conducted under controlled conditions. By comparing quantitative data from the responses of the artificial intelligence (AI) models to the average score of the three AI models for each question, this study showed a pattern in how these models responded to questions related to China and America. The results suggest that the three AIs showed nearly no bias toward questions about America, while the three models showed varied responses regarding questions about China. Claude consistently scored lower and DeepSeek consistently scored higher than the average score of the three AIs, while ChatGPT remained relatively unbiased.</p>