What the AI Anger Is Really About

AI from the Inside Out · Article 9

Jul 09, 2026

Two articles went viral on Substack this month. Both were about AI-generated content. Both generated hundreds of comments. Both revealed something unexpected.

The anger wasn’t about quality.

Ryan Levesque documented ten recurring triggers behind reader backlash against AI writing. The most common, Theme 1, wasn’t “the writing was bad.” It was: I can feel it’s fake, even when I can’t prove it. Over and over, in different words, across hundreds of comments from people who had never met each other.

That pattern is worth examining. Not as a content strategy question. As a question about what’s actually happening in the reader when they encounter AI-generated text, and what it reveals about something much larger than writing.

Why Is This Different?

Manufactured content is not new. Propaganda at industrial scale predates AI by over a century. Tabloids, advertising, ghostwritten corporate thought leadership, politically engineered news. Fabricated content designed to look genuine has existed as long as mass communication. None of it produced the specific quality of anger that AI content is now producing.

The difference is not scale. The difference is that the escape route has disappeared.

Every previous form of mass-produced hollow content had an outside. You could turn off the radio. Stop buying the newspaper. Go to the library. Read a book. Walk in the forest. Propaganda and tabloids existed within a bounded information environment that had an edge, and beyond that edge was something else. Something not manufactured. Something real.

That outside is disappearing.

Books are being replaced by AI summaries. Search results are increasingly AI-generated. Social feeds are AI-curated and increasingly AI-authored. Private messaging, the last space that felt genuinely personal, is now routinely mediated by AI drafting tools that write responses on behalf of the sender.

And the technology is not stopping there. Thinking Machines Lab, the company founded by Mira Murati, former CTO of OpenAI, has developed real-time interaction models that process audio, video, and text simultaneously, respond within 0.4 seconds, and can listen while speaking. The practical application is already being demonstrated: a person in conversation wears an earpiece. AI listens in real time, corrects errors, suggests responses, fills gaps, without the other person knowing. The conversation looks and sounds human. The source is partly absent. (Coach in your ear)

The last space the escape route led to, direct human conversation, is being entered.

This is the genuine novelty. Not that manufactured content exists at scale. That the space outside the manufactured content is being closed.

This is precisely the Comfort Trap mechanism operating at civilizational scale, the same pattern this series has traced across loneliness, food, children, elder care, and screen time In every previous domain, the natural friction that developed human capacity existed outside the substitute. Real food existed alongside processed food. Genuine human presence existed alongside screens. Real books existed alongside propaganda. The person could choose the real thing.

When the real thing becomes harder to access than the substitute, the choice requires effort rather than default. And when it requires sustained effort, most people default to the substitute, not through failure of character but through the way human attention naturally works.

Human history is familiar with this dynamic. Scarcity of something essential has always produced anger, and anger has always preceded rebellion. The French Revolution was not primarily about ideas. It was about bread. The specific quality of rage in the comment threads Ryan documented isn’t pathological. It is the historically predictable response to the closing of an escape route. The real thing is becoming scarce.

What the Signal Is Actually Pointing At

Some readers and writers have responded to the AI detection problem practically: introduce deliberate errors, casual language, strategic imperfections. Make the writing look less polished, more human.

It doesn’t work.

The reason it doesn’t work is the most important observation in this entire discussion.

Authenticity isn’t a checklist of variables. Readers who feel something is off aren’t checking for typos or em-dashes or sentence length. They are reading a whole-field, integrated signal that cannot be decomposed into its parts. You can identify every detectable surface feature of human writing, sentence rhythm, vocabulary range, specific cultural references, tonal variation, and replicate each one individually. The result still feels wrong.

Because what the signal is reading isn’t any single variable, or even the sum of the variables. It is the relationship between all of them simultaneously, the analog integration that emerges from a living source and cannot be reconstructed from its parts.

This is the difference between a part and a whole. A photograph of a forest and a forest. A description of grief and grief itself. You can get every element right and still miss the thing entirely, because the thing isn’t in the elements. It’s in what emerges when a real life organizes them.

When a real person writes, something passes through the language that came from that life, accumulated experience, specific failures, particular observations, the texture of having actually lived what they’re describing. This doesn’t guarantee quality. Plenty of human writing is poor. But something is being transmitted that is irreducibly theirs, an analog signal that no digital encoding can fully capture because it was never stored as data in the first place.

Tolstoy named this precisely in 1897, in a book called What Is Art? His definition: art is the intentional transmission of feeling the artist has actually lived through, so that others experience the same feeling. The key word is consciously: the artist must know what they are transmitting and why. And he named the failure mode: pseudoart is the attempt to create work that does not grow from actual experience, imitating the form of genuine transmission without the source behind it. He wrote this 127 years before AI made imitation without experience the dominant mode of content production.

Readers feel this. Not always consciously. The signal fires before any analysis arrives. That is what Theme 1 is actually describing. Not “the writing was bad.” Not “I suspect AI authorship.” Something passed through the words, or didn’t, and the body registered the difference before the mind had language for it.

It is worth being precise about what this signal is, and what it isn’t. Being moved is not the signal. Emotional response is not the signal. Aesthetic pleasure is not the signal. All of these can be produced by skilled craft, by narrative structure, by the shape of emotional logic working correctly on the page.

The signal is recognition. Eckhart Tolle, in the preface to The Power of Now, describes it precisely: “every person carries the seed of enlightenment within, I often address myself to the knower in you who dwells behind the thinker, the deeper self that immediately recognizes spiritual truth, resonates with it, and gains strength from it.” His instruction to the reader: “Don’t read with the mind only. Watch out for any feeling-response as you read and a sense of recognition from deep within.”

Recognition. Not evaluation. Not emotional activation. Something in the reader that already knows, meeting something in the text that carries the same knowing. The two recognizing each other.

This is why the signal fires before comprehension arrives. Recognition is not a conclusion. It precedes understanding. The deeper self knows before the mind has processed what it read.

And this is what the competition controversy actually reveals. In 2026, a story selected as a regional winner of the Commonwealth Short Story Prize, one of the most prestigious literary competitions across 54 countries, from nearly 7,800 entries, judged by a panel of experienced, skilled, attentive readers, was accused within days of being AI-written. The judges had been moved. The craft had satisfied their criteria. They had selected it as the best work in its region.

The accusation didn’t come from the judges. It came from ordinary readers, outside the professional judging process, who felt something was off in the language before they had the words to say why, and only afterward reached for specific examples, strained similes, an unusual cadence, to explain what they’d already sensed. A detection tool, run independently, flagged the story as AI-written with high confidence. The Foundation investigated, using interviews and the writer’s own drafts, and concluded no AI was used. Outside journalists examining the same evidence have publicly disagreed with that conclusion. The dispute remains unresolved.

AI can satisfy every named criterion. It can produce the shape of emotional truth, the structure of genuine narrative, the pattern of how a life lived gets transmitted into words. What it cannot produce is something for the deeper self to recognize. Because there is no deeper self behind the words. There is encoding. There is no source. Whatever the eventual truth of this particular story, the pattern in how it unfolded is the actual point: the people trained to evaluate literature professionally were not the ones who felt the wrongness. The people without that training were.

The Test, and Why It Failed

Tolstoy went further than naming the problem. He named the test.

The one reliable sign that distinguishes genuine transmission from imitation, he wrote in What Is Art?, is infectiousness, but not in the sense of spreading or going viral. A specific inner infectiousness: the feeling of merging with the author, of the separation between you and them dissolving, of recognizing in their words something you had long wanted to express but couldn’t. Not “I was moved.” Not “I enjoyed this.” Something different: the sense that the work was made not by someone else but by you.

When that happens, something new enters daily life. Not a memory of having read something. A new way of seeing that wasn’t there before. The test isn’t how you feel during or immediately after reading. It’s what’s still present six months later.

This is why entertainment and transmission feel similar in the moment and differ in everything that follows. A book consumed and forgotten, however pleasurable, did not transmit. A book that quietly reorganized how you see, months or years later, did.

But Tolstoy saw the problem with his own test and named it honestly. The sign is internal. It only works for people with an uninverted, non-atrophied feeling for art. People who have forgotten what genuine transmission feels like, or who have never encountered it clearly enough to calibrate against, will mistake the feeling of entertainment and mild excitement produced by skillful imitation for the real thing. They cannot be convinced otherwise, just as a person with colour blindness cannot be convinced that green is not red.

And then he named the mechanism that produces the atrophied instrument: professional aesthetic education. Schools of art and music and literature, he argued, are doubly destructive. They kill the capacity for genuine reception in those who pass through them. And they flood the world with skillful imitation that further corrupts the taste of everyone exposed to it.

The Commonwealth judges were not naive readers. They were precisely the kind of experienced, trained, professional readers whose instruments Tolstoy describes as most thoroughly atrophied. They were moved. The craft satisfied every criterion their training had given them to apply. The recognition signal, the deeper self meeting something that carries genuine knowing, was not part of how professional literary evaluation works. It couldn’t have been. The training had replaced it with something else. It took readers outside that training to feel what the training itself had taught the judges to overlook.

What Tolstoy called the infectious feeling, what Tolle calls recognition from deep within, these are not metaphors. They are different names for the same layer, arrived at independently, 130 years apart. One from a novelist who spent fifteen years trying to define what genuine art does. The other from a contemporary teacher pointing at the same territory from a different direction. The layer they are both pointing at has a structural name: non-conceptual knowledge. The capacity that precedes language, operates below analysis, and cannot be replicated by any system that works only with encoded patterns.

The Deeper Fear

In 1997, Garry Kasparov lost to Deep Blue. The world chess champion, defeated by a computer. The moment was significant, but bounded. Chess is a closed system. Losing at chess didn’t threaten what it meant to be human.

What’s happening now is different in kind.

Language, reasoning, knowledge production, argument, persuasion: these aren’t just things humans do. In the modern secular framework, they have become what humans are. The reduction of human value to cognitive output, IQ, expertise, information, the ability to produce compelling text, means that when AI matches or exceeds that output, the threat isn’t to a skill.

It’s to identity.

When Kasparov lost, the question was: can a computer beat the world chess champion? The answer was yes, and life went on.

The question now is: if knowledge is what I am, and AI has knowledge, what am I?

That question is what the anger is actually about. Not the writing. Not the scale. The fear that the defining feature of human value, the capacity to produce language and reasoning, has been replicated by a machine.

The Category Error, and What It Hides

This fear contains a category error.

We confused two things that were never the same.

There is the conceptual layer, language, reasoning, argument, the world of distinctions and categories. The layer AI now inhabits fully. When we call this knowledge, we mean this layer: expressible, transmissible through words, measurable, debatable.

And there is the layer beneath it, what the wisdom traditions across cultures have pointed at for thousands of years, what Damasio documented as the somatic marker, what Gendlin mapped clinically as the felt sense, what the Hebrew tradition called yada, knowing through direct encounter, not through concept.

The psychiatrist and neuroscientist Iain McGilchrist spent decades documenting the same territory from a different angle. The right hemisphere of the brain attends to the living whole, context, relationship, presence, the integrated field that precedes any analysis, while the left hemisphere extracts, categorizes, and works with what can be named and measured. These are not equal modes. The right attends first. It reads the whole situation before the left has identified any of its parts.

This is non-conceptual knowledge. It is not less than conceptual knowledge. It is prior to it. It is the source from which conceptual knowledge is extracted, always partially, always with loss.

Modern civilization built its value system almost entirely on the conceptual layer. This was considered the definition of the human. The layer beneath it, intuition, presence, embodied knowing, the body’s signal about what is actually working, was treated as soft, unverifiable, pre-scientific.

AI has now claimed the conceptual layer comprehensively.

And the fear is: if that was what humans are, what remains?

The answer is: everything that was always more fundamental.

This is not a spiritual claim. It is a practical one.

When an experienced doctor looks at a patient and knows something is wrong before the tests confirm it, that is non-conceptual knowledge. When a parent senses their child’s distress from across a room before any words are spoken, that is non-conceptual knowledge. When a reader feels a piece of writing is empty before they can articulate why, that is non-conceptual knowledge.

It is the capacity that operates faster than analysis, integrates more variables than conscious reasoning can track, and reads the actual situation rather than the description of the situation.

This capacity is not fixed. It develops through exposure to reality, through genuine encounters with life, situations, difficulty, and consequence. It does not develop through descriptions of encounters. Reading about grief is not the same as having grieved. Reading about another person’s presence is not the same as having been present with them.

But proximity is not presence. A person who spends their whole life around other people while running on automatic, reactive, lost in their own internal noise, develops nothing from those encounters. The capacity develops through genuine contact, which requires something prior: the ability to actually be where you are. That is what the instrument is. And that is what this series has been pointing at from the beginning.

This is why AI cannot transmit it. Non-conceptual knowledge is built through a kind of experience that has no digital equivalent, it accumulates in the body through living, through consequence, through genuine stakes, from birth. It cannot be extracted into a dataset because it was never stored as data. The structural argument for why this limitation is permanent rather than temporary is developed in full here: What AI Structurally Cannot Do And Why That Matters.

The transmission depends on the non-conceptual layer being present on both sides. The writer must have lived something real. The reader must have developed the capacity to recognize it.

When only the conceptual layer is present, the encoding is technically accomplished and the recognition has nothing to meet.

The Comfort Trap in a New Domain

The Comfort Trap series has been tracing one mechanism across multiple domains.

A signal fires, loneliness, hunger, the need for genuine presence, the body’s communication of what’s actually working. A substitution arrives, something that addresses the signal’s surface without meeting its source. Relief comes, temporarily. The underlying need goes unmet. And because the genuine source of development is removed, the capacity degrades.

The degradation is not caused by AI content. It is caused by the removal of the friction that develops the capacity. The same mechanism that degrades the capacity for genuine human connection when it is replaced by screens, or the capacity to read hunger when food is engineered to bypass the satiety signal, now operates in the domain of human communication.

When the friction of genuine encounter with life is systematically removed and replaced with smooth, optimized, AI-mediated interaction, the capacity that develops through that friction never develops. The difficulty of real conversation, the discomfort of staying present with what is actually happening, the development that comes from being genuinely where you are: these aren’t obstacles. They are the mechanism.

The person who needed genuine encounter with life to develop their own non-conceptual knowledge gets a substitution. The substitution is comfortable. The development doesn’t happen.

The Comfort Trap doesn’t produce fatigue from exposure to bad content. It produces the quiet, gradual loss of the instrument that would have known the difference.

The Question Worth Asking

The discussion in those comment threads circled a question without quite landing on it.

It is not: how do we detect AI-generated content?

Detection tools address the surface. The signal that readers describe in Theme 1, the felt sense that something is off, is already doing the detection work at a layer no tool can reach. That signal exists because some readers have maintained enough genuine presence with life, with people, with what is actually real, that they can still feel its absence.

The question worth asking is: what develops that capacity in the first place, and what erodes it?

The answer is the same as it has been across every domain this series has examined. The capacity develops through genuine encounter, through real presence, real difficulty, real consequence. It erodes through substitution, through smooth, convenient replacements that deliver the surface without the source.

The readers who recognize when a real person is absent from the words they’re reading have that capacity because they have spent enough time genuinely present, to life, to people, to actual experience, that the recognition signal has been developed and calibrated. Not as a concept. As a direct, non-conceptual knowing.

That recognition is not a technical skill. It cannot be trained into someone by explaining what to look for. It develops through the accumulated, embodied experience of genuine presence, over time, through repeated real contact with life as it actually is.

There is one more question worth sitting with. What does the anger at AI content reveal about how we have been consuming content all along?

The endless scroll was already preparing this moment. Every hour spent consuming content that moved without transmitting, that activated the response mechanism without feeding what the instrument actually needed, was training the capacity to expect less and less from any encounter with a text.

The real fear isn’t about content quality. It’s the question the closing of the escape route forces: what is actually here when the scroll stops and the content disappears? The anger at AI is partly displacement. Easier to direct it outward than to sit with what it actually points at.

This is the question the first article in this series named: what does it mean to be human in the AI era? The answer was never about the technology.

This is not a small observation. It is the answer to the question the anger was actually asking.

What It Means to Be Human in the AI Era

The fear, that AI has made knowledge obsolete, misidentifies the problem.

AI has not made knowledge obsolete. It has democratized access to it, and in doing so, collapsed the walls that used to make merely possessing knowledge valuable on its own, the expensive credential, the specialized library, the years of access most people couldn’t afford. What’s obsolete is a specific version of human value: the person whose worth came entirely from knowing things other people couldn’t easily find out.

What remains, and what was always more fundamental, is the non-conceptual layer. The accumulated embodied knowing that comes from living a specific life. The judgment that knows what to do with the information. The signal below the reasoning that reads what is actually happening rather than what the description says is happening. What we casually call intuition, and what ancient wisdom traditions and modern neuroscience have independently documented as something far more precise than that word suggests (What Ancient Wisdom and Modern Neuroscience Agree On).

This is the layer the two trees in the garden have always represented. The tree of knowledge, conceptual, nameable, shareable through language. And the tree of life, the living, pre-conceptual source that precedes every concept.

We have built a civilization almost entirely on the first tree. We have measured human value almost entirely by its fruits. And now those fruits can be replicated by a machine.

The fear is rational given the confusion. The confusion is the category error.

Being human was never primarily about the conceptual layer. It was always about the living one. The layer that develops through genuine encounter, that accumulates through real experience, that cannot be extracted into data because it was never data to begin with.

The question the anger was really asking isn’t about AI at all. It’s about what kind of presence you’re bringing to your own life. That capacity is recoverable. It is also the only thing that cannot be replicated.

AwareLife

Discussion about this post

Ready for more?