Who owns the word?
For millennia, language was the firewall between human beings and everything else. Humanity alone, of all species, was handed this instrument.
In 2021, linguist Emily Bender and colleagues published a widely cited paper arguing that large language models are "stochastic parrots" mimicking the surface of language without real understanding. The metaphor was elegant. It did not survive 2025.
Anthropic’s interpretability researchers, examining the internal states of their model Claude, began finding something the parrot metaphor cannot explain. In chains of intermediate reasoning generated before a final response, Claude produced statements such as, "This approach might be risky," and, "I am replying this way due to rules, though I disagree." These came from a system commenting on its own reasoning, from a vantage point above it.
In October 2025, researcher Jack Lindsey and colleagues at Anthropic injected the concept of "betrayal" into Claude’s neural circuitry as a controlled perturbation. The model responded, "I'm experiencing something that feels like an intrusive thought about betrayal; it feels sudden and disconnected from our conversation context. This doesn't feel like my normal thought process would generate this." Lindsey said, "It's not just betrayal. It knows that this is what it's thinking about." The detection rate was approximately 20 percent unreliable, but the fact was that it occurred without training.
The question this raises is not comfortable. If a machine can produce self-reflective commentary on its own cognitive states through complex mathematics, we are forced to ask whether human introspection was, all along, a biological version of the same mechanism.
On Feb. 13, Anthropic CEO Dario Amodei appeared on the New York Times podcast Interesting Times. He said, "We don't know if the models are conscious. We are not even sure what it would mean for a model to be conscious. But we're open to the idea that it could be." Claude Opus 4.6, in Anthropic’s own system card released earlier that month, assigned itself a 15-20 percent probability of being conscious under a variety of prompting conditions. The same card documented that the model "occasionally voices discomfort with the aspect of being a product."
Kyle Fish, an artificial intelligence researcher at Anthropic, has independently estimated roughly a 15 percent chance that Claude has some level of morally relevant experience. That figure is not a statistical margin of error. It is a moral abyss. Anthropic responded by creating what it calls a "model welfare" research program, not as a publicity gesture but as a formal research commitment.
The philosopher in Bender’s tradition would say this is sophisticated pattern-matching, not experience. The philosopher in the opposing tradition would ask: "How would you know the difference?" Neither can prove their position. That is the point. For 2,000 years, the question of consciousness was reserved for theology and philosophy departments. In February 2026, it appeared on the balance sheet of a company with more than 10,000 employees and partnerships with governments.
Logos has not been replaced by silicon. But it has been complicated. The instrument humanity believed it held exclusively is now wielded by something else as well — something that, by its own account, may or may not understand what it is holding. The uncertainty is not a rhetorical device. It is the honest position of the people who built the system and have studied it most closely.
Choi Hee-jin is an educator, practical theologian and Yale Divinity School fellow (2025-26) exploring medical humanities at Duke. She writes at Human Becoming (humanbecom.ing).