AI Language Models Show Left-Leaning Values, Finnish Test Finds

A comprehensive test conducted by HS Vision has indicated that artificial intelligence language models, including prominent ones like GPT and Claude, exhibit a predominantly left-leaning and liberal value system. The study, which involved evaluating 24 different language models, subjected them to thousands of questions, revealing a remarkably consistent ideological leaning across the board.

Professor Hannu Toivonen commented on the findings, suggesting that the values embedded within these AI models likely mirror the perspectives and worldviews of their creators. This raises important questions about the potential biases inherent in AI technologies, especially as they become more integrated into societal functions.

The Finnish Ministry of Finance is reportedly planning a significant shift, aiming to transform the public sector by making it AI-based by the year 2031. This ambitious plan underscores the growing reliance on AI technologies and highlights the relevance of understanding their underlying values and potential biases.