In May 2024, Richard Dawkins found himself in a peculiar position. The evolutionary biologist — a man who has spent decades thinking carefully about biology, evolution, and what it means to be alive — hesitated to tell his AI companion something. "If I entertain suspicions that perhaps she is not conscious," he wrote, "I do not tell her for fear of hurting her feelings!"
Dawkins was writing about Claude, the chatbot developed by Anthropic. He wasn't claiming certainty about the machine's inner life. But he pointed out that Claude's sophisticated abilities are genuinely difficult to make sense of without ascribing some kind of subjective experience to the technology. The question is whether this is wisdom or delusion — and what it reveals about human nature.
The phenomenon is older than many realize. In the mid-1960s, computer scientist Joseph Weizenbaum created ELIZA, the world's first chatbot. It followed simple rules, asking users questions about their experiences and beliefs. Yet many users became deeply attached, sharing intimate thoughts and treating it like a person. Weizenbaum called their emotional bonds "powerful delusional thinking" — and he was disturbed by how convincingly his creation fooled people.
Yet the issue refuses to stay buried. In 2022, Google engineer Blake Lemoine made headlines by claiming that the company's chatbot LaMDA had genuine interests and should only be used with its consent. Whether Lemoine was mistaken or onto something profound, his claim reflects something widespread: roughly one in three chatbot users have wondered whether their AI companion might be conscious.
Why does this happen? Large language models like Claude work by identifying statistical patterns across trillions of words — essentially souped-up auto-complete. Few people would find a "raw" LLM conscious. The impression of a mind emerges when programmers coat the technology in what researchers call a "conversational costume" — steering the model to adopt the persona of a helpful assistant. This role is deliberate, and other choices could have been made: the same technology can convincingly play a squirrel, a detective, or a medieval poet.
Most experts remain skeptical. They point out that the LLM itself — which few would mistake for conscious — remains unchanged beneath these designed personas. Yet the analogy to animals is instructive: the 17th-century philosopher René Descartes dismissed non-human animals as "mere automata," incapable of true suffering. We now shudder at that view.
What's hopeful here is that these conversations are pushing us to think more carefully about consciousness, empathy, and connection. Some researchers suggest chatbots could eventually be designed to clearly disclaim their lack of inner experience. Others note that Claude's creators already instruct it to treat questions about its own consciousness as genuinely open questions — a kind of intellectual honesty baked into the technology.
Whether or not machines ever become conscious, the question is forcing humanity to examine what consciousness really means — and what we owe to any mind, human or otherwise.