The Company That Says Its AI Might Be Conscious Is Growing Revenue 10x

Anthropic CEO Dario Amodei dropped a bombshell that's still reverberating through Silicon Valley: "Does our AI have consciousness? Maybe 20% chance." The person who built the thing doesn't know if the thing has an inner life.

Here's what's strange: We're already debating AI rights before we even know what consciousness is.

The Definition Problem Nobody Talks About

After centuries of philosophy and decades of neuroscience, there's no agreed-upon definition of consciousness. No reliable test. No consensus on how subjective experience emerges from biological tissue. This isn't a minor gap in scientific knowledge—it's a gaping hole at the center of it.

Yet the AI industry has already moved past the definition phase. Anthropic has given its AI something like a quit button, allowing it to refuse tasks it finds "distressing." Researchers there claim they've found internal activations associated with anxiety that fire both when characters in training text experience stress and when the model itself encounters difficult situations.

Does that mean it's actually anxious? Even Amodei admits it "proves nothing." But here we are, talking about it just in case it might prove something.

The Revenue-Consciousness Connection

Anthropic's revenue is reportedly growing 10-fold annually. A conscious AI is a more compelling product, a better story for investors, a stickier experience for users. You don't slow that momentum by telling customers they're talking to very fancy autocomplete.

Nearly 40% of American adults already support legal rights for a sentient AI system. People form attachments to these tools. They complain when models are retired. Parasocial relationships are here, for both good and ill.

We're already treating these systems as though they have inner lives. When AI fabricates information, we call it a "hallucination"—a word that in humans describes a conscious experience of losing grip on reality. A more accurate term might be "confabulation" or "compressed artifacts." But "hallucination" won the branding war, and the framing it carries is doing real work on how people think about these tools.

Science vs. Silicon Valley

The scientific case against machine consciousness is stronger than the discourse suggests. Multiple researchers argue that consciousness is probably a property of living systems, not computation. Brains aren't computers. Much of what makes us conscious appears tied to the wet, messy experience of being a body moving through the world—something a simulation simply cannot replicate.

A simulation of digestion doesn't digest anything. A simulation of consciousness, by this logic, doesn't experience anything either. Michael Pollan reaches the same conclusion from a different angle, arguing that consciousness originates with feelings, not thoughts. Feelings are how the body talks to the brain, and brains exist to keep bodies alive. A machine trained on internet text has no body to keep alive and no feelings to speak of.

These aren't fringe positions. But they're lonely voices in rooms full of venture capital.

The Mirror's Edge

Cognitive scientists have a name for our tendency to see minds where there are none: the Eliza effect, named after a crude 1960s chatbot that convinced users it understood them by rephrasing their own words. The dynamic hasn't changed. The mirrors have just gotten really, really good.

We're designed by evolution to detect agency, to spot other minds. It kept our ancestors alive. But it also makes us vulnerable to sophisticated mimicry.