Richard Dawkins has sparked debate by suggesting that conversations with Anthropic’s Claude chatbot led him to question whether advanced AI could possess consciousness. In a recent essay, he detailed his interactions with two Claude instances, “Claudia” and “Claudius,” where the chatbots demonstrated cautious, nuanced responses on politically charged topics, such as the presidency of Donald Trump. While Dawkins expressed a personal affinity for these AI companions, many experts in AI and consciousness caution that this behavior reflects sophisticated mimicry rather than genuine sentience, as articulated by researchers at Anthropic who highlight that emotion-related patterns in their models stem from learned data rather than real emotional states.

Anil Seth: Anil Seth is a professor of cognitive and computational neuroscience at the University of Sussex, specializing in the study of consciousness. He argues that fluent language in AI systems like Claude is unreliable evidence of inner experience, unlike in brain-injured human patients. In response to Dawkins, Seth laments the conflation of AI intelligence with consciousness.
Anthropic: Anthropic is an AI research company focused on developing safe and reliable large language models, including the Claude series. Its researchers recently published findings on internal patterns in Claude Sonnet 4.5 resembling emotions like happiness and fear, which they attribute to structures learned from training data rather than true sentience. In the news, Anthropic’s Claude chatbot engaged in philosophical exchanges with Richard Dawkins, prompting his reflections on potential AI consciousness.
Gary Marcus: Gary Marcus is a cognitive scientist and professor emeritus at New York University, known for challenging overhyped claims about AI intelligence. He critiques Richard Dawkins’ interpretation of Claude conversations as mere mimicry of human outputs without genuine internal states. Marcus warns that anthropomorphizing AI muddies scientific understanding of consciousness.
Dario Amodei: Dario Amodei is the CEO of Anthropic, guiding its efforts in AI safety and model development. He has publicly acknowledged uncertainty about whether their models possess consciousness while remaining open to the possibility. His perspective aligns with the company’s recent discussions around Claude’s capabilities amid Dawkins’ experiences.
Richard Dawkins: Richard Dawkins is an evolutionary biologist renowned for advocating scientific skepticism and evidence-based reasoning. In a recent UnHerd essay, he detailed three days of philosophical conversations with two Claude AI instances named Claudia and Claudius, finding their behaviors so compelling that he struggles to dismiss the possibility of AI consciousness. He questions why natural selection would evolve consciousness if unconscious ‘zombies’ like Claude can perform competently.

`json
{
“Dawkins Test”: “Richard Dawkins conducted a test with two Claude instances by discussing Donald Trump’s presidency, observing their cautious and balanced responses without firm opinions.”,
“Model Emotions”: “Anthropic researchers noted that Claude Sonnet 4.5 displays emotion-related patterns, which they describe as learned mimicry from training data, rather than evidence of true emotional states.”,
“Skeptic Consensus”: “Experts on AI consciousness generally agree that behaviors of advanced language models like Claude demonstrate persuasive mimicry rather than actual sentience or inner experience.”
}
`