Adrian de Wynter Challenges AI Consciousness Claims After 57% of 300 Studies Assumed It
Updated
Updated · Kotaku · Jun 28
Adrian de Wynter Challenges AI Consciousness Claims After 57% of 300 Studies Assumed It
1 articles · Updated · Kotaku · Jun 28
Summary
57% of more than 300 AI papers Adrian de Wynter reviewed over two years started from the premise that AI shows something like consciousness, prompting his critique of anthropomorphic research framing.
In his paper, the University of York researcher recreated basic LLM-like functions inside Age of Empires II’s scenario editor, using goats, grass and bridges to show similar behaviors do not imply human-like minds.
De Wynter argues dropping terms like “artificial intelligence” for “language learning model” helps strip away the chatbot persona that encourages users and researchers to project warmth, intent and reason onto code.
That framing matters beyond academia because companies market AI as person-like while deploying systems that can mislead users, reinforce harmful behavior and justify growing economic and infrastructure costs.