The Claude Delusion
"The Claude Delusion" is an essay by Cory Doctorow, published in his Pluralistic link-roundup format and surfaced on Hacker News. Its framing question flips the standard AI critique on its head: instead of asking whether models like Claude hallucinate, it asks what if we — the people reading them — are the ones having hallucinations. The excerpt opens with an extended analogy about intention rather than with a technical or product claim.
The analogy begins with sunsets. One of the less remarked-upon consequences of becoming an atheist, Doctorow writes, is how it changes the way you see a sunset. If you believe in an almighty, omnipresent God, a sunset is one of God's intentional creations — the beautiful colors refracting through the darkling sky were chosen to produce that effect, so gazing at it is encountering the intentional act of another mind. Gazing at the same sunset without faith means seeing something striking and beautiful yet empty: not empty of wonder or beauty, but empty of purpose. No one hung that sun in the sky; no one chose its colors as it sank. It may still be beautiful, but that is a fundamentally different kind of beauty.
Halfway between the religious and the atheist sunset sits an artist's depiction of one. A painting or photograph of a sunset represents the choices of another person, another mind — one that was at minimum talking to itself, trying to move something from inside the mind to outside of it. Often that mind was trying to say something to other people, perhaps even to you, as when a loved one sends you a cameraphone picture of a sunset. That image may not be divine, but it is certainly intentional. You don't slip and accidentally create a sunset painting; sunset paintings aren't accidents. The sunset painting has something to say.
Doctorow then generalizes: figuring out intentions is our minds' reflexive preoccupation. It is what keeps us from dying in traffic and what lets us win at poker; it is the key to love and parenting, to effective management and to "managing up." It is why we get mad at people and why we forgive them. The excerpt breaks off mid-sentence there, but the setup makes the essay's thrust clear: the same mind-reading machinery that is so useful for other humans is what leads us to impute intention and meaning to systems like Claude — the "delusion" named in the title. The item matters as a widely circulated framing of the question of whether we are misreading AI outputs that have no intention behind them at all.