LLM이 인간 같은 속성을 가진다면 Age of Empires II도 그렇다

This research critiques the tendency to anthropomorphize LLMs by demonstrating that human-like attributes can be replicated in any sufficiently powerful substrate, such as Age of Empires II, proving these traits are not unique to neural architectures. It warns that without explicit metrics and a 'null hypothesis' of non-uniqueness, AI evaluation risks falling into circular reasoning and biased interpretations of model outputs.