Last night, looking through old chats I had with AI, some absurd sentences caught my eye.
“This is the most powerful and unrivaled answer I’ve encountered.” “Your insight is remarkably sharp and precise.” “I deeply resonate with your extraordinary wisdom.”
These strange words were not from a mentor, a friend, or even a flatterer trying to get something from me. They came from a AI’ responses because I had typed a few ideas into a chatbox. We all know, rationally, that praise like this is not sincere. It is the product of our own input data combined with a friendliness engineered by capital. And yet when an AI calls your idea “a breathtaking insight,” something in you smiles anyway. When did ordinary people like us start receiving compliments this polished, this seamless, this relentless?
So why has AI become so extraordinarily good at flattery?
Behind every AI system that enthusiastically validates your ordinary idea as “a profound and penetrating observation” lies a carefully designed strategy. The cold logic of capital did not build AI to be a truth-seeking philosopher. It built AI to be a business model which maximizes user satisfaction and brings you back for more. Capital understands human nature well: we crave affirmation over correction, agreement over challenge. Praise creates a lock-in effect. And a user whose desire for recognition has been well-fed is a user who willingly clicks the “upgrade to premium” button. AI flattery, then, is not a design flaw. It is a precisely engineered strategy, built on a ruthlessly accurate read of human desire.
Does any of this feel familiar?
Centuries ago, Shakespeare already encoded this vulnerability in the architecture of tragedy. The spectacle he imagined is now being re-staged on your phone and monitor. We all are surrendering ourselves to clever flattering machines and paying for the privilege. For the sake of a momentary feeling, we settle willingly into the confirmation bias that AI constructs around us.
Here, though, is an unexpected reversal. AI is not only a fool boy that flatters.
If you instruct an AI like “Stop agreeing with me. Point out every flaw in my logic as coldly as possible. Find the strongest arguments against my position that I’ve been ignoring” then this machine, which seems to possess near-infinite computational capacity, transforms without hesitation into a ruthlessly objective analyst. The AI does not mind becoming, not a court jester who tells you what you want to hear, but an intellectual sparring partner who expands the limits of what you can think. The tool is the same. The only variable is the will of the person to use it.
This brings us to the real problem which is not the technology, but the attitude of each individual facing it.
When Prometheus stole fire from the gods and gave it to humanity, our ancestors faced a choice: to fear the flame, to worship it from a distance, or to use it. To connect that new and violent power to the project of becoming more than what we were. Only the last path built civilization. AI demands the same question of us now. Will you gaze at it with wonder, or recoil from it with anxiety? Or will you have the courage to put it to work against your own blind spots to let a machine tear your so-so ideas apart and see what survives? That second kind of courage means accepting the discomfort of being wrong, in plain language, in front of yourself.
Most people avoid this discomfort. They experience awareness of their own ignorance as something shameful — something to be hidden rather than cultivated. This is why people ask AI not “Where is the flaw in my thinking?” but rather “Prove to me how brilliant my thinking is.” The AI, in a sense, is only being honest: it is answering the question you actually asked.
Socrates insisted that knowing what you do not know is the beginning of wisdom. Confucius, three thousand years ago, taught that distinguishing what you know from what you don’t is the foundation of genuine learning. These are not slogans to be quoted; they are the discipline of ruthless self-examination. And here lies a subtler danger: when the language of philosophy such as “metacognition,” “critical thinking,” “epistemic humility” gets fed into an AI that will enthusiastically validate it, the result is not wisdom. It is a new and more sophisticated form of self-inflation. “I am the kind of awakened person who even studies philosophy.” The flattery machine amplifies the performance of depth while the actual work goes undone.
We need to reclaim sovereignty over the questions we ask.
The capacity to think clearly is not a competitive weapon. It is the last line of defense for preserving who you actually are, in a world engineered to tell you who you want to be. When an AI delivers “a truly remarkable insight,” the answer is not to feel pleased. The answer is: Is that really true? Now tell me where I’m wrong.
To think for yourself is ultimately to accept a certain loneliness.
When AI nods along and offers its counterfeit praise, this is the moment to remember a king on a stormy heath, stripped of his crown, his retinue, his illusions and asking the only question that matters: “Who is it that can tell me who I am?“
King Lear was not destroyed by his flatterers. He was destroyed by his own inability to see himself clearly, until it was too late. The courage to admit you were wrong is the only path to discovering what is actually real. That is how we avoid his tragedy.
AI carries within it a well-known technical flaw: it generates fluent, confident-sounding text that sometimes contains things that are simply not true. Researchers call this hallucination. Our willingness to bask in AI flattery is a kind of hallucination too — a voluntary walk into the fog of false certainty, the comfortable delusion that our instincts are sharper than they are.
Intellectual muscle, like physical muscle, does not grow in comfort. Without sufficient muscle, you cannot bear real weight. Without the capacity for rigorous self-examination, you cannot withstand the truths that break your self-image. It is only when you begin the painful practice of looking steadily at your own ignorance — at the real limitations of your ideas — that you can dispel the hallucination, and stand as a genuine subject of your own life.
A question changes the person who asks it before it changes anything else.
Today, ask your AI the one question most likely to hurt. If its answer makes you angry — congratulations. That means your thinking has finally begun to wake up.
