A Microsoft AI researcher has constructed a neural network using goats from the video game Age of Empires II, in a project that mocks the notion of chatbot consciousness. The so-called "goaty experiment" aims to highlight how readily people anthropomorphize large language models such as ChatGPT, Claude, and Gemini.
The researcher argues that the tendency to attribute human-like awareness to AI systems is an absurd leap, given that these models are fundamentally pattern-matching algorithms. By building a neural network from in-game goat units, the project underscores the artificiality of such claims.
No specific performance metrics or data were released. The experiment is described as a conceptual demonstration rather than a functional AI system, making direct comparisons to existing chatbots impossible.
This critique arrives as public debate intensifies over AI sentience and regulation. While anthropomorphizing chatbots may seem harmless, it could shape policy decisions and public expectations in misleading ways.
Critics might argue that the experiment, while clever, risks trivializing legitimate progress in AI research and ignores genuine advancements in natural language understanding.