The panic arrives in a familiar costume: the machine is getting too close to the sacred interior of human thought.
This time the worry is not that AI will take our jobs, flatten art, flood the internet, or make children cheat on essays. Those anxieties are still around, of course, but the newer version cuts deeper. AI, we are told, may be occupying the space where judgment used to live. It writes before we have struggled, summarizes before we have read, and offers the shape of an answer before our own uncertainty has had a chance to do its work.
TIME’s recent essay, “Are We Losing Our Minds to AI?”, gathers the research behind that fear and gives it a memorable name: cognitive surrender. The concern is real enough. If you let the machine frame the problem before you have even found your own question, you can end up mistaking fluency for thought. But the conclusion many people want to draw from this is far too simple. AI does not make people surrender their minds. It gives weak thinking a smoother place to hide.
“I’m just intentional about it, and I always try to think first and then prompt.”
Steven Shaw, quoted in TIME
That is the whole argument, hiding in plain sight.
The research does not say that AI use ruins thought. It says sequence matters. Researchers at the University of Chicago and University of Toronto found what they called a “temporal reversal”: under time pressure, early AI access helped; with enough time, late or no AI access produced better critical-thinking performance. The problem was not the presence of AI. The problem was inviting the answer into the room before the thinker had entered it.
That distinction matters because it changes the practical question. A panic story asks, “Is AI replacing our minds?” A better story asks, “Where in the workflow does thinking actually happen?” Those are not the same question, and collapsing them produces the very cognitive laziness the panic claims to oppose.
The Workflow Is the Argument
From a Dramatica perspective, the confusion is obvious. The Objective Story is about a new technology entering shared work, education, medicine, and culture. The Main Character Throughline is about agency: whether an individual still experiences judgment as something exercised from the inside.
The TIME piece keeps brushing against that distinction, then letting the atmosphere of dread blur it. AI becomes a kind of external antagonist, when the actual conflict sits in the relationship between tool, habit, and responsibility. The machine is not the whole source of conflict. The workflow is.
Anthropic’s large-scale interview study makes the same point in a less theatrical way. People reported both learning gains and cognitive atrophy concerns, but the pattern was uneven. Students and teachers raised the atrophy concern more often; tradespeople reported strong learning benefits with far fewer atrophy concerns. That is not a story about AI melting every mind it touches. It is a story about context, incentive, and whether the user is trying to learn or trying to escape the conditions of learning.
The difference shows up immediately in writing. A writer who asks AI for an argument before forming one has not gained a collaborator. They have gained a substitute for the first act of thinking. But a writer who brings a position to the model, asks for pressure against it, tests the counterargument, rejects weak frames, and returns to the page with sharper intent is doing something else entirely.
The tool did not remove the writer’s agency. It revealed where agency was missing.
Strong Tools Expose Weak Thinking
Packy McCormick’s older argument about AI and human cognition holds up better than the panic cycle around it. In “Evolving Minds,” he points to chess and Go as examples of machines exceeding human performance and then becoming part of the training environment that helped humans see new possibilities.
“AlphaGo made me realize that I must study Go more.”
Lee Sedol, quoted in Not Boring
That is the productive response to superior intelligence. The machine does not end the game. It changes the conditions under which mastery develops. Chess engines did not end chess. AlphaGo did not end Go. In both cases, the machine exceeded human performance and then became part of the environment through which human players learned new patterns, new possibilities, and new standards.
The danger was never that the engine could think. The danger was forgetting that mastery still requires the player to play.
That is where the “losing our minds” frame becomes misleading. It treats cognition like a possession that can be stolen, when thought is closer to a practice that can be strengthened, neglected, distorted, or trained. A person who asks AI for an answer to avoid the discomfort of forming one is not a victim of cognitive surrender. They are participating in it.
A person who argues with the model, tests its assumptions, rejects its framing, and uses it after forming an initial position is not surrendering thought. They are adding resistance. The same system can become a crutch or a sparring partner depending on whether the writer arrives with any stance of their own.
The Storyform Still Belongs to the Writer
For writers, the first private articulation of the problem matters. The blank page matters. The bad first version matters. The confused note in the margin matters because it proves the writer is still inside the work, still discovering what the pressure is, still trying to decide what the story means before polishing the expression of that meaning.
If you skip that step, the model’s first frame becomes your borrowed Storyform. You may still produce something coherent, but coherence is not ownership. A clean paragraph can hide the fact that the writer never decided what the paragraph was for.
Dramatica has always been useful here because it separates Storyform from storytelling. Storyform is the underlying argument of the story: how conflict is organized across Throughlines, Storypoints, Dynamics, and resolution. Storytelling is the expression of that argument: scenes, images, dialogue, rhythm, silence, and all the choices that make one writer’s work feel distinct from anyone else’s.
AI becomes dangerous when it is allowed to confuse those layers. It can generate storytelling before the Storyform has been chosen. It can produce tone before the writer has clarified pressure. It can make the surface feel finished before the underlying argument has become accountable.
That is not a reason to reject AI. It is a reason to place it correctly.
Use it after you have asked what the conflict is really about. Use it after you have written the ugly version. Use it after you can tell whether the model’s answer strengthens the story’s argument or merely decorates the page. Use it when it makes you more responsible for meaning, not less.
Think First. Prompt Second.
So no, we are not losing our minds to AI. We are discovering how many workflows were already designed to avoid thinking.
AI makes that avoidance faster, cheaper, and more polite. It can turn hesitation into output before the hesitation has taught us anything. It can let us outsource the very struggle that would have formed judgment. That is worth worrying about, but it is not worth mythologizing.
The answer is not abstinence. It is authorship.
Think first. Prompt second. Argue with the output. Keep the parts of the work that form judgment close enough to feel the friction. Let the machine widen the search space, pressure-test the claim, surface the counterargument, and accelerate the tedious pieces. But do not hand it the job of deciding what you mean before you have tried to mean something.
That is how you keep your mind.
Not by protecting it from tools, but by refusing to let the tool become the Main Character.