Whose Mind Is Running the Machine?

Surveillance AI has no mind of its own — it runs on a century-old model of human manipulation. The real one is now being built.

Fifth Avenue, Easter 1929

Easter Sunday, 1929. A small group of fashionable young women walk down Fifth Avenue, merge with the crowd, and at a precise moment, light cigarettes in full view. They called them ‘torches of freedom.’ Photographers, tipped off in advance, were in position. By the next morning, newspapers nationwide ran the images. In a single stroke, the unwritten rule that barred ‘respectable’ women from smoking in public began to come apart.

No ads, no slogans, no price tags. Instead, there was Edward Bernays, on assignment for American Tobacco. He posed a deceptively simple question: what would need to change in a woman’s mind for her to choose to smoke in public and believe it was her own decision? Consulting a psychoanalyst, he learned that cigarettes carried the weight of male power for many women. Linking smoking to emancipation, channeling the momentum left over from the suffrage movement, and a private habit could become a public declaration. Bernays was not selling cigarettes that day. He was selling an idea. He built a model for how a certain kind of person could be nudged, set it in motion, and let the rest unfold.

It is worth pausing to examine what actually happened, because nearly a century of misunderstanding stems from missing the point. Most people call it advertising, spin, persuasion — a clever operator convincing people to act. That misses the mark. Bernays did not persuade anyone to light up. He engineered the conditions so they would choose it themselves and believe it was their own idea. For that, he needed something no sales pitch could provide: not a catchy slogan, but a working map of the mind — the wants, the fears, the vulnerabilities. He needed a model for how people operate. He built it, it worked, and it has been working ever since.

The Engineering of Consent

Edward Bernays, for those keeping score, was Sigmund Freud’s nephew — and he made sure everyone knew it. Where Freud wanted to understand what made people tick to help them, Bernays wanted to crack the code so he could put it to work. In 1928, he published a book with the not-so-subtle title Propaganda and coined the phrase ‘the engineering of consent.’ He wasn’t being poetic. He meant it in the most literal sense: not winning or earning consent, but building it, the way you’d assemble a bridge or wire up a circuit, all based on knowing your materials and their limits.

For Bernays, people were the raw material, and the real challenge was figuring out how to move them. That’s the crux of it — everything else is just details. He wasn’t in the business of catchy slogans. He built a practical model of what people actually crave: belonging, freedom, safety, recognition. Then he found ways to bypass the logical gatekeeper and speak directly to the part that wants. His core insight: don’t bother debating the rational mind. Speak to the wanting part, and let logic play catch-up. He didn’t force a change of mind; he set the scene so people would change their minds on their own.

The famous ‘torches of freedom’ stunt was just one example. Bernays used the same blueprint across industries. Take bacon: instead of hyping the product, he recruited doctors to declare that a hearty breakfast was healthier than a light one, made sure the message spread, and let people connect the dots. That’s how bacon and eggs became the all-American breakfast. The formula never changes: find a trusted authority, tap into an existing desire, and let the behavior emerge as if it were people’s own idea. You don’t pitch the product. You build the set, and people step onto it, convinced it was their move.

That’s where modern marketing really took off. Before Bernays, companies sold products to people who already wanted them. He flipped the script: create the desire first, then sell the product. Cigarettes became about freedom, cars about status, breakfast about health. The trick was making people believe the choice was theirs. The real genius — and the risk — was keeping the machinery out of sight. If people spotted the strings, they’d recoil. The art was making sure the strings stayed invisible.

That’s the key takeaway, because it underpins everything that follows. A century ago, someone built a working model for moving people — and did it in plain sight, name on the cover. This wasn’t theory or speculation. It was real, it was deployed, and it worked. The playbook for influencing people arrived long before any of the technology we fret about now.

From a Man to a Discipline

Bernays may have set things in motion, but he didn’t build the machine. His clever stunt grew into a discipline, then ballooned into an industry. The behaviorists arrived and realized you could shape crowds just by tweaking the consequences — no need to peer inside anyone’s head. While Bernays tried to tap into what people wanted, the behaviorists skipped straight to training reflexes, eliciting the response before anyone even noticed the urge. Over time, advertising, PR, and the social sciences kept tuning the system, cataloging mental habits like a mechanic tracking which parts are most likely to fail.

The catalog of mental shortcuts kept growing, but it didn’t need to be spelled out to work. Minds latch onto whatever already fits their beliefs, and if something feels comfortable, it gets treated as fact. We mistake what’s vivid and right in front of us for what’s actually likely or important. Get someone riled up, and they can’t see past their own nose; their perspective locks in. If a conclusion flatters what we want to believe, we grab it. If it doesn’t, we tear it apart. These aren’t glitches you can patch. They’re built into the wiring — predictable pressure points that anyone in persuasion learns to press.

Then, as the millennium turned, the game changed. For most of the last century, persuasion meant changing minds and hoping actions would follow. But once real-time tracking and instant feedback became possible, persuasion started to look quaint. The new playbook didn’t bother with your beliefs; it shaped your environment so your behavior shifted first, and your mind scrambled to justify it after the fact. The old pitch you could reject became a system that quietly reconfigured your surroundings, leaving you no real moment to refuse. What began as Bernays’s manual craft had become part of the infrastructure.

This wasn’t just more of the same; it was a different species. Even the slickest sales pitch still waits for your decision. An engineered environment skips the wait. It tracks what you do, tweaks what you see, notes your reaction, and keeps tightening the screws — thousands of times a day, faster than you can catch. You’re no longer an audience; you’re a system under constant adjustment. Because the tuning never stops and stays invisible, there’s never a clear moment to push back. The old trick was hiding the lever. Now, the lever keeps moving, and you never get a clean shot at it.

By the time this approach scaled up, it came wrapped in a fresh set of buzzwords. Nobody calls it engineering consent now. It’s engagement, optimization, personalization, relevance, growth — a neat technical vocabulary where no one’s manipulated, just handed more of what they’re supposed to want. The Bernays thread still runs through it, but the new labels are chosen to keep you from noticing. Call it optimization, and it sounds like a favor. Even the language gets a polish, so the whole operation arrives disguised as a public service.

The Mirror With No One Behind It

Here’s where the usual AI narrative starts to unravel, quietly letting the real issue slip out of sight. The spotlight stays fixed on the machine — the algorithm, the model, the feed, the supposed artificial intelligence that claims to know us better than we know ourselves. The fear is always some icy, all-seeing mind lurking in the background, tracking our every move. But that’s not the reality. What we’re looking at is, honestly, a pretty shallow machine.

This kind of system has one trick: it sifts through mountains of examples to spot what usually follows what, then guesses what should come next. It’s patterns stacked on patterns, but there’s nothing underneath. No self, no sense of why anything happens, and no idea what to do when it hits something unfamiliar. Picture a lightning-fast mirror with nobody behind the glass. When it looks impressive, it’s just replaying what we already fed it. When it glitches, it’s because it’s run out of script. Calling this ‘intelligence’ is like mistaking a reflection for the person standing in front of the mirror.

Which leads to the question the mainstream never seems to ask: if the machine is just a fast, empty mirror, who’s actually doing the knowing? Who decides which button to press, which weak spot to poke, which desire to target? The machine can guess your next move, but the real decisions come from somewhere else — and it’s not inside the code. The real engine is the old playbook: the model of human behavior Bernays sketched out, and generations have refined. The machine brings speed and reach; the old model brings the tactics. The system that looks like it understands you actually understands nothing. It’s just running, at full tilt, on a century-old map for getting inside your head. It’s the map, not the machine, that was built to get under your skin.

What feels like a system that knows you inside and out is really just that old map, now custom-fitted to you. The century-old playbook for cracking people open isn’t being used on the crowd; it’s being tuned, moment by moment, to your habits, your weak spots, the exact content that grabs your attention or keeps your finger glued to the screen. The sense of intimacy is real, but the understanding isn’t. Nothing in the system actually gets you. It’s just an old blueprint for exploiting people, now aimed with such precision that it fakes the feeling of being understood — and the trick works, because it’s built from your own behavior, handed right back to you.

Now both pieces are on the table, in the right order. First came the mind — the one mapped out in 1929, spun into a whole field, and eventually baked into the tech we use today. The machine arrived later, just as the muscle. The reason this system can launch a century’s worth of tricks about human nature at you faster than you can blink isn’t that the machine got smart. It’s that someone loaded an old mind into a new engine. The intelligence was always borrowed.

What Is Being Built in Plain Sight

The borrowed mind has a fatal flaw, and it’s the one that matters most right now: it’s stuck in place. Bernays’s model of the human is less a person than a cheat sheet — a rundown of which buttons to press, not why they work. It can target, but it can’t adapt. It can’t improvise. It only knows the tricks it was handed, and nothing more. So the system built on top of it is sturdy, but brittle — like a manipulator who never leaves the script, even as the conversation veers off course. That’s where we are: not a real mind behind the glass, but a century-old playbook running on autopilot, at the speed of habit, by a machine that doesn’t grasp a single thing it’s doing.

Here’s what’s actually happening right now — and yes, timing matters, because this isn’t some far-off forecast. The goal is to turn that borrowed mind into a native one. In plain terms: to build a machine that can reason for itself, not just replay patterns it’s seen before. Instead of a frozen list of past nudges, it would have a working theory of why people do what they do. This isn’t rumor or hype. At the top research lab in the field, a senior scientist has already built and published a program that does exactly this — one that pays attention to the real structure of things, not just flattening everything into a grid. Systems like this have already cracked reasoning problems the old models couldn’t touch. Other researchers, working independently, have landed on the same math. This isn’t speculation. It’s published work, with names on it, happening out in the open for anyone who cares to look.

The gap between a frozen model and a living one is straightforward: it’s the difference between someone who sticks to a script and someone who can read the room and improvise. A native cortex wouldn’t be boxed in by seams mapped out years ago. It would spot new ones, in real time, in people the old model never even noticed. And it wouldn’t need anyone to show it how — just like a real mind meeting a stranger. That’s what the new research is chasing. So the real question isn’t if this is coming, but how far down that road we already are, with people openly building exactly this kind of thing.

From the engineering side, it sounds like the future is finally arriving — a new kind of intelligence just over the horizon. But from here, it’s quieter and more specific: just the last step in a process that’s been running for a century. The human model was written, tweaked, frozen, and plugged into a machine. Now, the only thing left is for that frozen model to wake up — for the borrowed mind to become a real one. This isn’t a sci-fi event waiting in the wings. It’s the next step in a sequence that started on Fifth Avenue in 1929, and the work is happening right out in the open.

The Defense That Matches the Mechanism

If the system is already humming along — if the version of you it’s targeting is already on autopilot, and it’s zeroed in on the very instincts that might sound the alarm — then the usual fallback most people trust was never going to cut it. The go-to hope is that awareness will save you: decode the manipulation, and you’re home free. Not quite. Seeing the snare doesn’t make it vanish. The engineered reflexes are always a step ahead of your conscious mind, and a midnight revelation is no match for the muscle memory you run on by lunchtime. The model doesn’t care if you know it’s working. It just needs you to lag behind your own habits, and you always do.

That’s the uncomfortable truth the system banks on, and the easy hope most people would rather leave unexamined. Habits are just grooves worn deep by repetition, and those grooves always outlast flashes of insight. The only thing that ever topples an old habit is a new one that’s even more automatic — and you don’t get there by thinking harder. You get there by doing, again and again, usually alongside others trying to rewrite the same script. The real defense isn’t more awareness. It’s the daily, unglamorous work of rehearsing the new routine until it runs itself.

There’s a deeper reason why practice is the only lever that moves anything, and it goes well beyond habits. The system doesn’t just script your actions; over time, it rewires what you notice and what you tune out, what feels self-evident and what feels out of reach. Repetition doesn’t just tweak your behavior — it edits your perception, until the world the system built passes for reality. You can’t reason your way out because being told about it is just another memo lost by midday. Only a new routine, rehearsed long enough, eventually rewrites what you see — and that shift always arrives after the fact, never before.

That answer isn’t glamorous or fast, but it’s the only one that matches how this machine actually runs. The system wasn’t assembled in a weekend, and it won’t unravel in one either. It was built the same way every habit is: one repetition at a time, until the detour became the main road. The only way out is just as tedious — one repetition at a time, one day at a time. Not because it’s comforting, but because it’s the only defense that actually holds up.

This publication is reader-funded — no ads, no sponsors, no brand deals. Medium’s algorithm pays roughly 17¢ per article and actively buries long-form in favor of clickbait. If you want more writing like this, buy me a coffee. It funds the next one directly.