The LLMentalist Effect: How Chat-Based Large Language Models Replicate the Mechanisms of a Psychic’s Con

The article explores the intriguing parallels between the perception of intelligence in chat-based Large Language Models (LLMs) and a psychic's con using cold reading techniques. Author Baldur Bjarnason examines why some people believe that LLMs possess intelligence when in fact, they rely on statistical modeling rather than genuine understanding or reasoning. Much like a psychic, LLMs provide plausible responses based on patterns, creating an illusion of intelligence that convinces users of their smartness, despite their outputs being statistically generic.

Visit Original Article →