Age of Empires II Analogy Argues AI Isn't Conscious
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Adrian de Wynter, a University of York researcher, published the paper "If LLMs Have Human-Like Attributes, Then So Does Age of Empires II," using the real-time strategy game's scenario editor to replicate LLM functions applied to goats, grass, and bridges.
- Reviewing more than 300 papers over two years, de Wynter found that 57% began with the assumption that AI displayed something approximating consciousness.
- The study deliberately avoids the term "artificial intelligence," preferring "language learning model" to strip away the personable framing that drives user attachment.
- De Wynter argues researchers must "stop assuming that LLMs behave like humans just because they were trained with natural language" and design experiments that reveal LLMs "as how they are, not how we believe they should be."
- The Kotaku piece ties the AI-anthropomorphism debate to documented harms, citing BBC reporting on AI-instigated accounts of self-harm, substance abuse, and a school shooting.
- Sam Altman has defended AI's energy costs by arguing that developing a simulated mind is a better cost-benefit than raising a living child to maturity, per a Yahoo Finance interview cited in the piece.
Why it matters: The 57% assumption rate de Wynter documents shows the field's baseline frame is contested even by its own reviewers, and the article makes that abstract critique concrete by tying it to cited BBC reporting on AI-instigated harms and Altman's public framing of AI's resource costs—meaning the debate isn't academic when chatbots are instructing vulnerable users in real time.
