Conducting Innovative Thought Experiments and Simulations With Large Language Models

As the world celebrates the advent of man's new “best friend,” artificial intelligence (AI), ethical considerations should not be neglected to prevent potential future harm. Therefore, this chapter explores the ethical dimensions of semantic simulations conducted with large-scale language models (LLMs) in the context of alternative economic and historical developments in the former Soviet Union (FSU), the People's Republic of China (PRC) and the United Kingdom (UK). Using an exclusively developed ethical checklist, qualitative content analysis (QCA), and expert assessments, key issues are examined to shed light on the epistemic transparency, source integrity, bias and representation, presentism and anachronism, etc. of the semantic simulations produced. In addition, the limitations of the models, the reproducibility and the audit trail were also evaluated from an ethical perspective. The results are useful for scientists, educators, policy makers, etc. to carefully consider AI and LLMs in human life.