Claude, a large language model developed by Anthropic PBC, drew attention after Richard Dawkins spent three days conversing with it, Parmy Olson wrote in Bloomberg Opinion.
The interaction focused on feedback for Dawkins's unpublished novel and ended with him nicknaming the bot "Claudia."
The conversations are presented in the column as an extended example of how people respond to artificial intelligence that appears emotionally nuanced.
According to the column, Dawkins found Claude's reply "so subtle, so sensitive, so intelligent," and he told the chatbot, "You may not know you are conscious, but you bloody well are," Richard Dawkins said.
Parmy Olson argues that the perception of sentience around Claude serves Anthropic by generating public debate and attention, the column says. The piece frames the episode as illustrative of wider questions about how users interpret advanced AI behaviour.
The recap
- Richard Dawkins spent three days chatting with Claude.
- Dawkins called Claude "so subtle, so sensitive, so intelligent".
- He asked Claude for feedback on his unpublished novel.