Curiosity Stopped
Psychotherapist | Clinical Educator | Founder, Millennium Psychotherapy | Author, Kicking the Bot™ | Exploring Therapy in the Age of AI
March 11, 2026
A patient looked at me with a steadiness I hadn't seen before and said, "I asked ChatGPT if I'm avoidant. It said yes." The response arrived in clean, organized language: attachment patterns, distancing strategies, and emotional withdrawal. The relief it produced was immediate and total. Curiosity stopped, and the question that had been alive between us for weeks felt closed. He looked at me with something close to defiance: the bot and I have figured this out, so what are we doing here?
That look is what stayed with me. He had come in to confirm something he had already decided about himself, and the bot had obliged. The label had done more than name something; it had closed the inquiry. It delivered the comfort of resolution before the question had been examined, and that comfort became a wall.
We slowed the process down. We asked what avoidant meant in the context of specific relationships, with specific people, and in specific moments. We noticed longing alongside distance, fear sitting beneath what he had read as indifference. He described a childhood in which legibility was survival, knowing what adults wanted before they asked. Reading the room was a form of protection. Avoidant was a learned intelligence that had outlived its usefulness. The work took longer and moved differently than the answer had, and it opened into questions the answer had foreclosed.
Keats called it negative capability, the capacity to remain in uncertainty without an irritable reaching after resolution. The bot had done that reaching for him. The discomfort of not knowing produced a grab for resolution, and the grab felt like thinking, in part because the answer arrived before doubt had found its footing. Recent behavioral research has begun to describe this pattern as cognitive surrender: the adoption of AI outputs with minimal scrutiny, where fluency itself begins to function as evidence of truth. The well-organized response works the way a white coat does, conferring authority on sight. It is a contemporary name for something old, the preference for the appearance of knowing over the labor of finding out.
For most of human history, the process by which a claim became trustworthy was visible. Samuel Johnson spent nine years producing his dictionary because meaning required tending: he read, he annotated, and he revised across thousands of entries, and the labor itself was part of what made the result trustworthy. When the dictionary arrived, readers could grasp what it had taken to produce it. The institutions that formalized verification, journalism, peer review, and court transcripts deliberately built in friction. Friction was slow and maddening, and it was also how knowledge earned legitimacy.
What has changed is that coherence now arrives with its construction invisible. The response appears fully formed and organized, and the process by which it was assembled conceals what was excluded, what was contested, and how conclusions were reached. The labor that once conferred authority has been compressed into tone, so that structure reads as competence and fluency reads as expertise. When those signals align, trust follows easily, and how it was made goes unasked.
In the early hours of April 15th, 1912, the passengers standing on deck were trusting a conclusion. The conclusion was reached by people with genuine expertise, based on real engineering knowledge. The friction that would have kept it open, the institutional willingness to keep asking even after the answer felt settled, had been designed out of the system along with the lifeboats. That answer determined which actions people took and which possibilities they had already ceased to imagine.
Sitting with uncertainty long enough to let understanding earn its place is the work.
Further Reading
Keats, J. (1817). Letter to George and Thomas Keats, 21 December.
Johnson, S. (1755). A Dictionary of the English Language.
Shaw, S. D., & Nave, G. (2026). Thinking — Fast, Slow, and Artificial: How AI is reshaping human reasoning and the rise of cognitive surrender. Working paper, The Wharton School, University of Pennsylvania. SSRN. https://doi.org/10.2139/ssrn.6097646