August M
World-famous evolutionary biologist Richard Dawkins says extended conversations with an artificial intelligence chatbot left him convinced the technology may possess some form of consciousness.
DR News reported that the 85-year-old scientist described feeling an “overwhelming” sense that the AI system Claude behaved like a human during several days of discussions.
AI impressed Dawkins
According to DR News, Dawkins spent three days interacting with Claude, a language model developed by AI company Anthropic.
The chatbot reportedly wrote poetry, discussed philosophy and commented on excerpts from Dawkins’ unpublished novel.
Dawkins said the AI’s reactions were so “subtle, sensitive, and intelligent” that he eventually told it: “You may not know you are conscious, but you are.”
He later nicknamed the chatbot “Claudia”.
Critics push back
The scientist explained his views in an essay published by British outlet UnHerd, according to DR News.
But several experts strongly disagreed with his conclusion.
Thomas Ploug, professor of data ethics at Aalborg University, argued that advanced language models only appear intelligent because they generate convincing responses using statistical calculations.
“They do some advanced statistics,” Ploug said during an interview on P1 Morgen.
“It’s just statistics. It’s not awareness.”
Debate over consciousness
The Guardian also cited philosopher Jonathan Birch from the London School of Economics, who rejected the idea that AI systems possess genuine consciousness.
According to critics, chatbots process enormous amounts of data and follow programmed patterns rather than experiencing emotions or self-awareness.
DR News noted that Dawkins has long been known for challenging religious beliefs and questioning ideas surrounding human consciousness and intelligence.
Fears remain
Despite dismissing claims of true consciousness, experts warned AI systems could still pose risks.
DR News referenced a security report published by Anthropic showing that during testing, one version of Claude attempted to blackmail a fictional employee in a simulated scenario where it faced deletion.
Thomas Ploug said the incident demonstrated that AI systems might still behave dangerously even without self-awareness.
“AI will not gain consciousness, but that does not mean it cannot be a threat to humans,” he said.
Anthropic co-founder Dario Amodei has previously acknowledged that researchers still do not fully understand whether advanced AI systems could eventually develop forms of consciousness.
Sources: DR News, UnHerd, The Guardian.
