The LLMentalist Effect: how chat-based Large Language Models replicate the mechanisms of a psychic’s con
Summary
The author argues that LLMs do not truly possess intelligence, instead producing statistically plausible responses. He compares the user illusion of ‘AI’ to the psychic’s con, detailing mechanisms like the Forer effect, subjective validation, and RLHF. The piece urges caution for businesses and readers when evaluating generative AI claims and provides references.