I’ve spent the last few years thinking about intelligence, looking to understand human, artificial, and the strange space between them. Most of the debate is around capability. How fast can it reason? How accurately can the answer be? How convincingly can it write? But somewhere along the way, I began to sense that we were measuring the wrong thing. Intelligence is usually judged by performance, and even our admiration of large language models follows that pattern that, when unleashed, ends up with the curious notion of artificial general intelligence. Said simply, we are impressed by what these models can produce.
Yet the most meaningful moments of human intelligence in my own life have had very little to do with production. They have had everything to do with transformation. And this led me to a deceptively simple yet startling insight.
A human being can be changed by their own thinking.
I’ve seen it in medicine. A clinician pauses, reconsiders a decision, feels that faint unease that something doesn’t fit. That reconsideration is not just a correction—it carries responsibility. It may alter treatment, outcomes, even identity. The thought does not simply solve a problem; it leaves a mark.
I’ve seen it in parenting. A difficult recognition about one’s own impatience reshapes memory and redirects future behavior. That insight lingers, and it shapes the future.
Psychology has long described related phenomena. Cognitive dissonance captures the unease we feel when belief and behavior collide and the internal friction that follows. But let’s add a bit more context to this ubiquitous human experience. Philosopher L.A. Paul has described “transformative experiences,” moments that just don’t add information but alter who we are. In both cases, thought does not merely update data. It reshapes the thinker.
In these moments, intelligence is not just the ability to generate possibilities. It is the capacity to be altered by them, because the experience itself fundamentally transforms the person experiencing it.
Large language models display extraordinary computational ability. They move through “possibility space” with remarkable speed and capability. But the outputs do not, in any way, attach to an interior life. They don’t accumulate biography and are not reshaped by what they produce. Each response stands alone.
AI computation generates, while human cognition transforms.
In education and clinical settings, I’ve begun to notice something subtle but significant. When tools deliver instant, polished synthesis, people often seem less connected to the result. The answer feels correct and efficient, but sometimes less substantial. And as a result, retention may be shallower. None of this diminishes artificial intelligence’s ability to accelerate discovery or surface blind spots. But I wonder whether frictionless fluency occasionally allows us to bypass the very discomfort that makes insight sticky or durable. We need to understand that genuine reflection is rarely this smooth. It involves—no, incorporates—the bumps of hesitation and doubt. And this complex friction interacts with identity as it moves forward in time and context.
The mind is permeable to itself.
But in a culture dominated with a sort of cognitive instant gratification, something may be shifting. When reflection becomes externalized, we may drift into the precarious zone of selection rather than transformation.
For me, the concern isn’t that machines can think. It’s that we may become less willing to let thinking change us. And if intelligence includes the capacity for self-alteration, then preserving this in the age of artificial intelligence may depend on protecting our own “cognitive permeability” and our human capacity to be reshaped by our own reasoning.
In a world of reflection on demand, the rarest act may be allowing a thought to leave a mark.
