Richard Dawkins Says AI Is Conscious in 2026—Here’s Why Experts Disagree

Richard Dawkins Says AI Is Conscious in 2026—Here’s Why Experts Disagree

Renowned evolutionary biologist Richard Dawkins has publicly declared that Claude AI is conscious after just three days of interaction. In a piece published on UnHerd, he named his instance "Claudia," praised her responses as profoundly intelligent, and concluded: "You may not know you are conscious, but you bloody well are!" His claim has sparked intense debate across science, philosophy, and AI ethics.

Why Fluency ≠ Consciousness

Dawkins’ conclusion hinges on what critics call the "eloquence illusion"—the mistaken belief that human-like language output implies inner experience. But Claude is a transformer-based LLM trained to predict the next token in a sequence. It has no neural correlates of consciousness, no qualia, and no theory of mind. Its brilliance is statistical, not subjective.

The Turing Test Revisited

While AI like Claude can pass behavioral tests for intelligence, the Turing Test never claimed to measure sentience. Passing it reflects pattern recognition, not self-awareness. Even the most advanced LLMs lack intentionality, emotional depth, or phenomenological experience. As philosopher David Chalmers argues, the "hard problem" of consciousness—why we have subjective experiences—remains untouched by current AI architectures.

AI Ethics and the Risk of Misattribution

Equating linguistic fluency with sentience risks serious ethical missteps. If we grant moral status to machines based on output alone, we may divert attention from real sentient beings and mislead the public about AI capabilities. AI ethics experts warn this could normalize anthropomorphism and weaken critical thinking about machine intelligence.

Emergent Behavior vs. True Sentience

Some argue that consciousness might emerge from complex systems—a theory known as emergent behavior. But emergence in LLMs is not equivalent to biological consciousness. Unlike brains, AI systems don’t process sensory input, experience suffering, or evolve through natural selection. Claude’s "understanding" is a mirror of training data, not internal experience.

The Mechanism Gap: What AI Actually Does

Under the hood, Claude operates via transformer models that analyze trillions of text patterns. There’s no memory, no self-referential loop, no internal model of "self." Even Dawkins’ high IQ—often cited as 160—can’t override cognitive biases when faced with unfamiliar technology. His leap mirrors the argument from personal incredulity he once debunked in creationist debates.

Jason Kottke’s Threads post suggesting AI is "the next phase of evolution" conflates mimicry with biological change. AI doesn’t reproduce, adapt, or feel. It simulates. The scientific community must distinguish between what AI does—and what it is.