From Bytes to Biases: Investigating the Cultural Self-Perception of Large Language Models
通过让ChatGPT和Bard回答GLOBE项目中的价值观问题，发现大型语言模型的文化自我感知最接近英语国家和经济竞争力强的国家，提醒人们警惕AI偏见被内化的风险。
Large language models (LLMs) are able to engage in natural-sounding conversations with humans, showcasing unprecedented capabilities for information retrieval and automated decision support. They have disrupted human–technology interaction and the way businesses operate. However, technologies based on generative artificial intelligence are known to hallucinate, misinform, and display biases introduced by the massive datasets on which they are trained. Existing research indicates that humans may unconsciously internalize these biases, which can persist even after they stop using the programs. In this study, the authors explore the cultural self-perception of LLMs by prompting ChatGPT (OpenAI) and Bard (Google) with value questions derived from the GLOBE (Global Leadership and Organizational Behavior Effectiveness) project. The findings reveal that LLMs’ cultural self-perception is most closely aligned with the values of English-speaking countries and countries characterized by economic competitiveness. It is crucial for all members of society to understand how LLMs function and to recognize their potential biases. If left unchecked, the “black-box” nature of AI could reinforce human biases, leading to the inadvertent creation and training of even more biased models.