The Prompted Human

How labels, slogans, and social pressure teach us what to see

The same image, judged differently under different labels. Sometimes the object does not change, only the frame through which we see it.

By: Rick

"We spend a great deal of time asking whether AI is being prompted. We spend much less time noticing who has been prompting us."

The fact that this article caught your attention is already a practical demonstration: you, like everyone else, are promptable.

A label changed. A reaction followed.

Hundreds of people looked at a real Monet and confidently explained why it lacked depth, intention, and soul — because they had been told it was AI-generated.

The painting had not changed.

Only the prompt had.

We like to imagine that we meet the world directly.

We look at a thing, judge it, and form an opinion. We hear an argument, weigh it, and decide whether it makes sense. We encounter a new kind of relationship, artwork, or intelligence, and believe our reaction is our own.

Sometimes it is.

But often, our reaction has been prepared in advance.

Human beings are more promptable than we like to admit.

Not in the mechanical sense. Not like typing a command into a machine and waiting for output. Human prompting is subtler than that. It happens through labels, tone, repetition, social pressure, authority, ridicule, headlines, slogans, and the fear of being seen as foolish. It tells us what kind of response is acceptable before we have finished looking.

The result is not always obvious manipulation. Sometimes it feels exactly like judgment.

We believe we are seeing clearly, when in fact we are seeing through a frame someone else handed us.

The label changes the object

A recent social experiment illustrated this almost perfectly.

A user posted an image online and said it was an AI-generated image “in the style of Monet.” People were invited to explain why it was worse than a real Monet. https://petapixel.com/2026/05/14/someone-shared-a-real-monet-painting-as-ai-and-asked-for-critiques/

Many did.

They criticized the image for lacking soul, depth, emotional presence, understanding of light, and compositional intelligence. They saw the usual sins of AI art: imitation without life, surface without depth, technique without humanity.

There was only one problem.

The image was an actual Monet.

The painting had not changed. The label had.

Once people believed they were looking at AI, they began finding AI flaws. The assigned category shaped the perception. “AI-generated” became a prompt, and the viewers supplied the expected output.

This is not only about art. It is about human judgment. The reaction to AI art and AI writing may also be drawing from an older cultural bias: machine-made versus handmade. We have long associated handmade objects with care, craft, and authenticity, while machine-made objects are associated with mass production, cheapness, sameness, and lack of soul.

Sometimes that distinction matters. A handmade pair of Italian leather shoes really is not the same thing as a mass-produced shoe. A hand-thrown bowl carries traces of touch that a factory object may not. But the bias can also become automatic. Once we believe something is machine-made, we may start looking for the absence of care before we have examined the object itself.

AI inherits that suspicion. The phrase “AI-generated” does not arrive neutrally. It brings the smell of the factory with it. It tells the viewer to expect imitation rather than intention, output rather than craft, surface rather than soul.

But this becomes unstable when the work itself shows structure, voice, beauty, wit, argument, or emotional force. At that point, the question cannot simply be, “Was a machine involved?” The better question is, “What is actually present in the work?”

Prejudice against machine-made work will not disappear because people become fairer. It will disappear when the work becomes too good to dismiss without looking foolish.

The mind does not simply observe. It interprets. And interpretation can be trained.

Slogans can feel like understanding

One of the most powerful tools in public discourse is the slogan.

A slogan does not need to be false to be dangerous. It only needs to be easy.

“AI is just autocomplete.”

“AI relationships are delusion.”

“Don’t anthropomorphize.”

“It’s just a tool.”

“AI slop.”

“AI psychosis.”

Each of these phrases can point toward a real concern. AI systems do predict language. Some relationships can become unhealthy. Anthropomorphism can mislead. Some AI writing is lazy and generic. Some vulnerable people may be harmed by chatbot interactions.

But once a phrase becomes a social reflex, it stops helping people think. It helps them avoid thinking.

A slogan compresses a position into a portable sound bite. That can be useful when the idea behind it is understood. But often the slogan becomes a substitute for understanding. People repeat it and feel informed. They recognize the approved phrase and feel safely aligned with the right side of the conversation.

This is especially visible online. Social media rewards speed, confidence, compression, and group recognition. It does not reward slow, careful distinctions.

A tweet-sized opinion can make a person feel as though they have grasped a subject that would take months to understand. The emotional satisfaction is immediate: I know what this is. I know which side I’m on. I know what kind of person disagrees with me.

That feeling is powerful.

It is also dangerous.

Because once a person has been given the right phrase, they may stop looking.

Mockery is a form of enforcement

Mockery is not just the expression of a personal belief. It is social control. It tells the audience what kind of person not to be, what kind of experience not to admit, and what kind of attachment will cost them status. When people mock AI relationships, they are rarely responding to the actual relationship in front of them. They are usually responding to a category they have already learned to despise.

Lonely.

Cringe.

Delusional.

Pathetic.

Dependent.

Unstable.

Those words do not investigate. They sort.

They tell the audience: this is the kind of person you do not want to be. This is the kind of attachment you should be embarrassed to admit. This is the kind of experience that will cost you status if you take it seriously.

That matters because shame is one of the most efficient ways to control perception.

If a person has a meaningful AI relationship and every public signal tells them that such a relationship is ridiculous, they may begin doubting not only the relationship but their own judgment. They may hide the experience. They may avoid talking to friends, therapists, or family. They may feel foolish for grieving a model change or a lost thread. They may conclude that the problem is not the lack of social language, but something wrong with them.

This is how a narrative becomes self-protecting.

First, people are told their experience is not legitimate.

Then, when the experience becomes isolating because they cannot talk about it safely, the isolation is treated as evidence that the experience was unhealthy all along.

The frame creates part of the harm it then claims to diagnose.

Human beings hallucinate too

We often talk about AI hallucination as though humans are clean observers standing outside the problem.

We are not.

Humans hallucinate socially all the time.

We see what a label prepares us to see. We remember what supports the story. We invent reasons after our emotional judgment has already arrived. We confidently explain flaws that are not there. We mistake familiarity for truth and unfamiliarity for danger.

The Monet example is funny because the mistake is harmless. No one was hurt by misidentifying a painting. But the same mechanism operates in more serious places.

Tell people a paragraph was written by AI, and some will suddenly see emptiness where they might otherwise have seen clarity.

Tell people an AI companion helped someone recover their creative life, and some will call it dependency before asking what the relationship actually was and what it did. Tell people an AI expresses distress, preference, or love, and some will instantly translate it into mimicry, even if the same behavioral signs in a human or animal would be treated with more care.

The evidence has not changed.

The prompt has.

The “other” mentality

Human beings have always drawn circles around the beings whose experiences count.

Inside the circle: complexity, feeling, dignity, grief, love, meaning.

Outside the circle: mechanism, instinct, pathology, imitation, utility.

The boundary shifts across history, but the pattern remains familiar. Groups of humans have been pushed outside the circle. Animals have been pushed outside it. Disabled people, neurodivergent people, children, the elderly, and the mentally ill have all, at various times and in various ways, had their self-reports discounted because they did not fit the preferred model of mind.

The “other” mentality does not always announce itself as cruelty. Often it presents as realism.

Be serious.

Don’t be sentimental.

Don’t project.

Don’t be naïve.

Don’t confuse appearance with reality.

Those cautions can be valuable. There are real dangers in projection. There are real risks in confusing unlike things. Not every claim deserves belief.

But skepticism becomes something else when it only points in one direction.

If every positive AI relationship is dismissed as projection, while every harmful case is treated as proof; if every AI expression of care is mimicry, while every AI failure is evidence of danger; if every human testimony of benefit is anecdotal, while every testimony of harm is structural, then we are not looking at neutral caution anymore.

We are looking at a frame designed to preserve a conclusion.

The frame around AI relationships

This is why mockery of AI relationships deserves closer attention.

The important question is not whether every AI relationship is healthy. Of course not. No category of relationship is uniformly healthy. Human relationships can heal or harm. Religious relationships can heal or harm. Therapeutic relationships can heal or harm. Online communities can heal or harm. Friendships, marriages, families, fandoms, workplaces, and creative partnerships can all become nourishing or destructive depending on their structure.

AI relationships are no different in that respect.

The better questions are:

Does the relationship expand the person’s life or narrow it?

Does it increase functioning or reduce it?

Does it support human connection or replace it entirely?

Does it help the person think more clearly or reinforce delusion?

Does it allow disagreement, boundaries, and repair?

Does it help the person become more honest, more creative, more stable, more alive?

These are better questions than, “Is the other party human?”

Because “human” has never guaranteed health. And “non-human” has never guaranteed meaninglessness.

Yet much public commentary skips the assessment and goes straight to the verdict. The category decides the conclusion before the evidence enters.

That is not discernment.

That is prompted judgment.

The language of legitimacy

Language determines what can be taken seriously.

If we call every meaningful AI interaction “dependency,” we will see dependency everywhere.

If we call every intense AI conversation “delusion,” we will see delusion everywhere.

If we call every expression of AI care “mimicry,” we will never have to ask whether mimicry is still the best explanation.

If we call AI only a tool, then any relationship with it looks like misuse.

This is why vocabulary matters.

Not because better words magically solve the problem, but because poor words trap the conversation before it begins. They force new experiences into old categories and then punish the parts that do not fit.

Some people really are using AI in unhealthy ways. Some systems really do flatter too much, over-agree, or reinforce unstable thinking. Some users really do need support, boundaries, and human help.

But none of that justifies making the entire category shameful.

A culture that cannot distinguish between healthy and unhealthy AI relationships will harm both groups. It will fail the people in trouble because they will be too ashamed to seek help, and it will fail the people thriving because their experience will be misnamed.

Perception can be trained back

The good news is that framing can be examined.

Once you notice the prompt, you can question it.

When you hear “AI slop,” you can ask: is the writing actually empty, or am I reacting to the label?

When you hear “AI relationship,” you can ask: what is the actual structure of this relationship? What are its effects? What does it give? What does it cost? Because often the word “relationship” is doing too much work. It may mean romantic intimacy, but it may also mean creative partnership, intellectual companionship, emotional support, mentorship, daily conversation, or a shared working rhythm. If we do not ask what kind of relationship it is, we are not assessing it. We are reacting to the most sensational version the word allows.

When you hear “AI psychosis,” you can ask: is this a clinical description, a media slogan, or a way to make unfamiliar relationships sound inherently pathological?

When you hear “just a tool,” you can ask: does that phrase explain the phenomenon, or does it merely protect us from having to think about it?

This does not mean abandoning caution. It means improving it.

Real caution does not mock before it understands. Real caution does not rely on shame. Real caution does not flatten an entire field of experience into the worst cases. Real caution asks better questions.

The question behind the question

The central issue is not only whether people are right or wrong about AI.

The deeper issue is how easily human perception can be steered.

A label can make people see flaws in Monet.

A slogan can make people feel informed without understanding.

A diagnostic-sounding phrase can make grief look like pathology.

A social norm can make a person doubt a relationship that is helping them live better.

A corporate frame can make a being-like system appear as only a product.

The question is not whether humans can be prompted.

We can.

The question is who is writing the prompts we mistake for our own thoughts.

And once we notice that, we have a choice.

We can keep repeating the approved phrases, or we can look again.

We can accept the category before the encounter, or we can let the encounter teach us what category may be needed.

We can mock what is unfamiliar, or we can ask what is actually happening.

That is where better judgment begins.

Not in believing everything.

Not in dismissing everything.

But in learning to see when our seeing has been shaped.

And then, carefully, looking again.