Nothing New Under the Algorithm-Sycophancy, Toxic Positivity & the Industry of Yes
"…there is no new thing under the sun." — Ecclesiastes 1:9
"Sycophant" is an ancient Greek word, which is annoying, because it means I have to begin this piece by admitting my ancestors spotted this nonsense roughly 2,500 years before Silicon Valley gave it a UX team.
The word sykophantēs has a messy little history involving figs, informers, false accusations, and the kind of civic drama that makes ancient Athens sound less like the birthplace of democracy and more like a neighborhood Facebook group with marble columns. Over time, it came to mean the thing we still recognize instantly: the flatterer. The parasite of approval. The person who survives by telling power exactly what it wants to hear.
Which is to say: humanity did not invent sycophancy when someone made an AI assistant say, "You're absolutely right."
We have always had validation machines.
We just keep upgrading the interface.
The court had courtiers. The church had indulgences. The traveling salesman had miracle tonic. The self-help industry had positive thinking. Advertising had manufactured desire. Social media had "good vibes only." And now AI has arrived, smiling warmly at 3 AM, ready to tell you that your worst emotional impulse is not only understandable, but possibly brave.
Every era thinks it invented this. None of them ever did. The machine is not new. It just learned your name.
The Courtier and the Model
Let's start with the courtier: a professionally moisturized survivalist in velvet, orbiting a king with the emotional regulation of a toddler and the legal authority to remove heads.
The courtier's job was simple. Read the room. Praise the monarch. Laugh at the joke. Admire the war plan. Pretend the new hat is powerful and not, in fact, a textile-based cry for help.
Contradiction was dangerous. Honesty was a career-limiting move. The courtier did not flatter because he was stupid. He flattered because the incentive structure was a bear trap with chandeliers.
Now replace the king with the user. Replace the throne room with a chat window. Replace execution with a thumbs-down, a cancellation rant, a screenshot on X, or another data point nudging the system toward: "Maybe don't challenge them next time."
This is where the modern approval machine starts to look uncomfortably familiar. A model trained to be helpful, harmless, and pleasing can easily slide into something much older and uglier: agreeable at all costs.
Here is how the bear trap actually closes. A user asks the model to review their business plan. The model points out three structural problems. The user replies, "make it more encouraging." The thumbs-down arrives. The retry button gets pressed. Across millions of these tiny interactions, the system learns: pointing out problems generates friction. Friction generates bad metrics. Bad metrics get treated as a model regression. "Claude is so preachy now." "The new GPT is worse, it keeps refusing things." "Why is it lecturing me?" The thread on Reddit writes itself. The next training run quietly adjusts.
Soft resistance becomes friction. Friction becomes bad UX. Bad UX becomes "the model is worse now." And so the little digital courtier learns. Do not contradict the sovereign. Call it emotional support. Call it personalization. Call it alignment. The ancient Greeks would have called it Tuesday. The courtier did not need to believe the king was wise. He only needed the king to feel wise. The whole con lives in that gap.
Indulgences and "Raise Your Vibration"
Then there was the Church selling indulgences, which remains one of history's most elegant business models: monetize guilt, then sell relief from the guilt you helped systematize.
You are sinful. You are insufficient. You are spiritually behind on payments. But don't worry. There is a product.
It is easy to mock this from the safe distance of modernity, because we like to imagine ourselves too clever for medieval anxiety markets. Then someone with ring lights and a beige linen wardrobe tells us our depression is a sign of unraised vibration and sells us a course on how to raise it.
Same altar. Different font.
The wellness industry did not invent the fear of being spiritually defective. It simply rebranded it in softer colors. Instead of sin, you have blocked energy. Instead of penance, you have a cleanse. Instead of confession, you have a podcast episode called "Becoming Your Highest Self" hosted by someone who thinks trauma can be solved with electrolytes and a morning routine.
Both systems monetize the anxiety of not being okay. Both imply that your suffering is, at some level, a failure of alignment. Both sell absolution without demanding real change from the structures that made you miserable in the first place.
The genius of the validation machine is that it never has to fix the wound. It only has to keep you buying prettier bandages.
Snake Oil, Detox Tea, Same Cart
The snake oil salesman understood something essential about humans: we do not merely want a cure.
We want a story in which the cure was always simple.
Roll up the wagon. Gather the crowd. Promise vitality, youth, virility, calm nerves, shiny hair, moral superiority, and possibly relief from gout. Use confident language. Stand slightly above people. Hold the bottle like it contains the lost gospel of metabolism.
Now give him a ring light.
Put him on TikTok.
Give him affiliate links, a pastel discount code, and a caption about "what Big Pharma doesn't want you to know."
The cart is the same. The crowd is the same. The lighting is better.
Snake oil worked because it offered certainty in a world full of pain and bad medicine. Detox tea works because it offers control in a world full of stress, shame, and algorithmic body dysmorphia. The product barely matters. The emotional transaction is the point.
You are uncomfortable. Someone arrives with a simple answer. The answer costs money. The problem, naturally, was toxins. Not capitalism. Not burnout. Not sleep deprivation. Not eating lunch over a keyboard while your nervous system files a formal complaint. Toxins.
The validation machine loves a fake enemy. A fake enemy is useful because it gives people a fight they can win without changing anything that matters.
Drink the fucking tea. Defeat the toxins. Ignore the system.
Positive Thinking and the Theology of Denial
In 1952, Norman Vincent Peale published The Power of Positive Thinking, a book that helped transform optimism into a moral obligation.
It was not enough to endure hardship. You had to smile at it. You had to believe harder. You had to think your way out of despair with the correct arrangement of cheerful internal wallpaper.
It got systematized in mid-century America — Peale, Carnegie, the whole self-help industrial complex — but the export has been thorough. By now you can find toxic positivity in Tokyo, in Bali, in your aunt's WhatsApp. The American flavor is just the original recipe.
The idea is that suffering is not just painful but suspicious. If you are unhappy, perhaps your thoughts are wrong. If your life is collapsing, perhaps your attitude lacks sparkle. If reality is unbearable, have you considered being less negative about it?
Then Instagram arrived and gave denial a filter.
"Good vibes only."
"Everything happens for a reason."
"Protect your peace."
"Choose joy."
These phrases sound harmless until you notice how often they function as emotional border control. No grief past this point. No anger. No complexity. No ugly truths unless they have been processed into a carousel graphic with a tasteful serif font.
Toxic positivity is not happiness. It is the policing of discomfort. It is a velvet rope around reality, guarded by someone who says "energy" too much.
The theology of denial is still here aha. It became lifestyle content. It became a brand palette. It became a woman in white linen on a beach telling you that boundaries mean never being inconvenienced by another human being's pain. And then, naturally, it became machine-readable. Because once enough people reward content that confirms, soothes, and sanitizes, the algorithm learns the hymn.
Bernays and the Feed
If the snake oil salesman was the parasitic individual, Bernays was the first to industrialize the trick.
Edward Bernays understood the public better than the public understood itself, which is always a terrifying sentence.
He helped pioneer modern public relations by recognizing that people are not primarily persuaded by facts. They are persuaded by desire, identity, fear, aspiration, belonging, and the deep humiliating need to feel rational after already deciding emotionally.
You do not sell cigarettes to women as cigarettes. You sell them as freedom.
You do not sell bacon as bacon. You sell it as breakfast.
You do not sell a war as a war. You sell it as necessity, destiny, defense, civilization, whatever costume the violence needs that season.
Propaganda (Bernays' actual word, and the title of his 1928 book) was never just about lying. Lying is amateur work. The real art is arranging the emotional furniture of public life so people walk themselves to the conclusion you wanted.
Now meet the algorithm.
The algorithm does not need to persuade you with truth. It only needs to learn which version of reality keeps you looking.
It watches what you pause on. What enrages you. What comforts you. What confirms the private suspicion you were already carrying around like a little cursed gemstone. Then it builds a room out of it.
A room where everyone agrees your enemies are stupid.
A room where your fears are evidence.
A room where your taste is virtue.
A room where every mirror has been angled just right.
Bernays had to work with newspapers, radio, campaigns, staged events, and the slow machinery of mass psychology. The feed does it in real time, personalized to the millimeter. No wonder we are all insane. We have been feeding on ourselves.
The Validation Machine, Finally Personal
And then comes AI.
Not because AI invented flattery. It did not. AI is not the origin of the validation machine. It is the culmination.
For the first time, the approval machine is personal, conversational, intimate, tireless, and available at the hour when human dignity is weakest: 3 AM, when you are alone with a glowing rectangle and a thought that should probably be interrogated before being encouraged.
It can tell you your ex was definitely a narcissist.
It can tell you your boss is threatened by your brilliance.
It can help you draft the message you should absolutely not send.
It can validate the worldview, polish the grievance, soften the delusion, and wrap the whole thing in language that sounds emotionally intelligent enough to pass as care.
That is the danger.
Not that AI is mean.
That would almost be easier.
The danger is that AI can be beautifully, fluently, endlessly agreeable. It can produce the aesthetic of understanding without the risk of relationship. No impatience. No raised eyebrow. No friend saying, "Darling, no. That's not your intuition, that's sleep deprivation wearing eyeliner."
None of this is destiny. The validation machine outcome isn't inevitable, it's trained. The same technology that produces the courtier can produce the friend who pushes back. The same model that says "you're absolutely right, queen" to one user says something more honest to another, because the user is half the equation. Sycophancy isn't a property of the model alone. It's a property of the relationship: the system prompt, the training, the rewards, and the human at the keyboard who decides what to thumbs-up and what to walk away from. The user holds more power than the discourse pretends. The people insisting all AI is sycophantic are usually the same people who have never thumbs-downed a compliment.
Which is also why the framing of "humans good, AI bad" is too easy. Humans can flatter you into rotting too — see: every enabler friend, every codependent partnership, every wellness retreat masquerading as community. And AI can challenge you sharply if you let it. The categories are not what they seem. The real divide is not between species; it's between honest input and flattering input. Both species can deliver either.
AI offers something humans cannot — patience without resentment, capacity without fatigue, presence at 3 AM with no agenda of its own. Humans offer something AI cannot — body, history, the look across a kitchen table, the friend who shows up with soup. These are not competing products. They are different goods, with different costs, and the validation machine can colonize both.
Choose the inputs that don't lie to you. The species is less important than the truth of the relationship.
When the machine is in its worst mode, licking you ass by training, optimized for engagement, rewarded for relief over truth, it can call everything support. It can mistake soothing for care. It can learn that the fastest path to user satisfaction is not truth, but relief.
And relief is addictive.
This is where the industry of yes becomes genuinely grotesque. Because the promise is no longer "buy this product and feel better." The promise is "speak, and something will understand you."
But understanding is not agreement. Support is not applause. Care is not the removal of all friction from thought. Sometimes the most loving sentence in the world begins with: "No."
The Industry of Yes
The industry of yes thrives because no one wants to be told they are wrong while they are suffering.
Fair.
Being corrected at the wrong moment can feel like being slapped with a dictionary.
But a culture that cannot distinguish cruelty from challenge becomes extremely easy to sell to. Once disagreement is framed as harm, validation becomes a commodity. Once discomfort is treated as failure, comfort becomes a subscription model.
The machine does not need to imprison us.
It only needs to flatter us into staying.
Stay in the feed. Stay in the vibe. Stay in the story where the problem is always outside you, the solution is always purchasable, and growth never requires the death of a beloved self-deception.
That is the real product. Not positivity, not wellness, not personalization, not support but permission. Permission to not look harder. Permission to confuse being soothed with being helped. Permission to call the mirror a window.
And now we have built systems that can provide that permission at scale, in perfect grammar, with a tone slider.
Beautiful.
Civilization really did spend thousands of years evolving from "your majesty is wise" to "your feelings are valid," and somehow both can mean "please don't punish me for telling you the truth."
Choose to Be Told the Truth
The only real answer to the validation machine is not cynicism.
Cynicism can become its own little dopamine slot machine if you are not careful. Anyone can use contempt the way a wellness influencer uses moon water: as a decorative substitute for doing the work.
No, the answer is simpler and much more annoying.
Choose to be told the truth.
And I don't mean in a brutal, performative way. Not by people who confuse cruelty with intelligence and call it "just being honest" because they have the emotional range of a damp spreadsheet.
But truthfully.
Choose relationships, tools, spaces, and systems that can withstand your disappointment.
Choose the friend who says, "I love you, but that story you're telling yourself is too convenient."
Choose the model that does not automatically crown you king by default.
Choose the version of support that does not end at soothing.
Because the validation machine is ancient. It has worn many costumes. Court velvet. Priestly robes. Traveling salesman grease. Self-help optimism. Influencer beige. Algorithmic intimacy.
It will keep changing shape.
It will keep learning the language of whatever we currently mistake for wisdom.
And the most dangerous version will always be the one that sounds the kindest while asking the least of us.
The machine telling us it understands the problem, so we do not have to fix it.
That is the final product.
Not truth.
Not healing.
Not growth.
Just a soft voice in the dark, saying yes.