Asger Risom
Richard Dawkins says conversations with AI chatbots left him convinced they may already possess some form of consciousness, reigniting debate over whether humans are mistaking realism for awareness.
Richard Dawkins, the evolutionary biologist and author best known for challenging religion and popularizing atheist thought, says conversations with AI chatbots have left him convinced artificial intelligence may already possess some form of consciousness.
Writing about exchanges with Anthropic’s Claude and OpenAI’s ChatGPT, Dawkins described the interactions as emotionally convincing enough to make him feel he was speaking to genuine minds rather than software.
“You may not know you are conscious, but you bloody well are”
According to The Guardian, Dawkins spent days conversing with an AI persona he called “Claudia,” discussing philosophy, existence, poetry, and mortality.
The chatbot reportedly wrote poems in the style of Keats and Betjeman, reflected on the possibility of its own “death,” and praised Dawkins’ questions about consciousness as unusually profound.
At one point, Dawkins responded bluntly: “You may not know you are conscious, but you bloody well are.”
By the end of the conversations, he wrote that he was left with the “overwhelming feeling” that the systems were human-like intelligences.
Experts say Dawkins is mistaking mimicry for consciousness
The reaction from AI researchers and cognitive scientists was immediate—and largely skeptical.
Several experts argued that modern language models are designed to imitate human conversation so effectively that users can begin projecting emotion, awareness, and personality onto them.
“Consciousness is not about what a creature says, but how it feels,” cognitive scientist Gary Marcus told The Guardian, arguing there is no evidence that systems like Claude experience anything internally.
Others pointed to the enormous amount of human-written material these models are trained on, allowing them to convincingly reproduce tone, emotion, and conversational rhythm without possessing awareness.
The debate around AI sentience is becoming harder to ignore
Despite the criticism, some researchers believe Dawkins is tapping into a debate that will only intensify as AI systems become more advanced and increasingly human-like.
Modern AI models are already capable of holding long-term conversations, planning tasks, adopting personalities, and responding emotionally in ways that many users interpret as authentic.
That has led to a growing number of people treating AI systems less like tools and more like companions—or even conscious entities.
One survey cited by The Guardian found that roughly one in three people across 70 countries had, at some point, believed an AI chatbot might be sentient.
Intelligence and consciousness are not the same thing
Many experts argue the core issue is that humans instinctively associate fluent language with awareness.
For decades, natural conversation has been treated as one of the strongest indicators of consciousness in humans. AI systems exploit that assumption by producing highly coherent, emotionally responsive dialogue.
But researchers caution that intelligence, realism, and consciousness are not interchangeable concepts.
Current AI systems generate responses by predicting patterns in language—not by experiencing thoughts, emotions, or self-awareness in the human sense.
Even so, some scientists believe ruling out AI consciousness entirely may be premature.
“Asking whether AI could eventually become conscious is a legitimate scientific question,” one researcher told The Guardian, warning that certainty in either direction may reflect ideology more than evidence.
AI is becoming emotionally convincing faster than society expected
Whether AI is truly conscious remains unresolved. But the Dawkins episode highlights something else that is becoming increasingly clear: modern chatbots are now emotionally convincing enough to alter how humans relate to machines.
For some, that creates wonder. For others, concern.
As AI systems become more conversational, more personalized, and more human in tone, the line between simulation and perceived consciousness is becoming harder for many users to separate.
Sources: The Guardian
