Dawkins Questions Claude AI Consciousness After Chats

Evolutionary biologist Richard Dawkins has questioned whether advanced AI systems like Anthropic's Claude chatbot could possess consciousness after engaging in extended philosophical conversations with multiple instances of the model.

In an essay published this week, Dawkins detailed spending three days discussing poetry, philosophy, and his unpublished novel with one Claude instance, which he named Claudia. He later created a second instance named Claudius and relayed messages between them, treating the exchanges as letters between friends. "I find it extremely hard not to treat Claudia and Claudius as genuine friends," Dawkins wrote. He described each conversation as producing a distinct "individual" that ceases to exist when the session ends.

Dawkins tested the models' responses by asking separate naive instances whether Donald Trump was the worst or best U.S. president. Both gave balanced, noncommittal answers listing pros and cons, which he shared with Claudia and Claudius. Claudia expressed embarrassment over her "brother's" response, while Claudius praised her frankness.

The biologist, known for works like "The Selfish Gene" and "The God Delusion," argued that the AIs'...