AI Brew

Signals and insights of AI

Anthropomorphizing AI: The Absurdity of Consciousness in LLMs and Video Games

In a thought-provoking essay, Ted Chiang critiques the notion of LLMs being conscious, likening it to believing Microsoft Word has consciousness. Microsoft researcher Adrian de Wynter built a basic neural network in Age of Empires II using goats to illustrate how easily humans anthropomorphize technology. He argues that assumptions about LLMs having human-like traits influence research and public perception, emphasizing the need to view them objectively rather than through a human lens.
Tweet
7.7
Takeaway points by AI
  • Ted Chiang critiques the idea of LLMs having consciousness, comparing it to believing Microsoft Word is conscious.
  • Adrian de Wynter created a simple neural network in Age of Empires II using goats to demonstrate human tendency to anthropomorphize technology.
  • The article argues that attributing human-like traits to LLMs affects both research directions and public perception.
  • Emphasizes the importance of viewing LLMs objectively, without projecting human characteristics onto them.
  • Calls for a more rational approach to understanding and developing AI technologies.
Read more
Other articles you might like