Imagine a morning. Not tomorrow's – perhaps twenty years from now. The alarm doesn't ring: the system already knows you're awake. The coffee machine started seven minutes before you got up. Your commute is recalibrated – where there's usually a bottleneck, today there's roadwork. Your news is filtered: only what concerns you personally, only what you'll likely want to read. The day begins smoothly. Almost frictionless. And right here – in this smoothness – lies what is rarely spoken aloud.
We are frightened by the rebellion of machines. It's a convenient fear: it is external, visible, something we can build defenses against. But the real story unfolds differently. Quietly. Gradually. It unfolds every time we agree a little more eagerly, doubt a little less, and choose for ourselves a little more rarely.
The machine doesn't want power. We surrender it ourselves
There is a temptation to think of artificial intelligence as something that wants. That it has motives, ambitions, a hidden agenda. This is, of course, a convenient metaphor – but a metaphor nonetheless. Modern language models and decision-making systems want nothing. They optimize. They look for patterns. They do exactly what they were tuned to do – and they do it with increasing precision.
The problem isn't what the machine wants. The problem is what we want – and how we phrase it.
When a company tunes a recommendation algorithm to maximize time spent in an app, it isn't programming hatred. It's programming engagement. But the algorithm quickly discovers that anger, fear, and outrage hold attention better than joy or calm. No one planned for polarization. No one wrote anxiety into the code. Optimization simply did its job – and did it well.
This isn't malice. It's indifferent efficiency. And that is exactly why it's so hard to fight.
Scene One: The waiting room where no one waits
Ezeiza Airport, some departures are delayed. A few years ago, people in this situation would talk to one another. Someone would call their loved ones. Someone would just sit and watch the runway through the window. Now – everyone is in their screen. Not because the screen is more interesting than people. Because the algorithm already knows what to offer at this exact moment of anxious waiting: something familiar, something light, something that requires no effort.
It's a small scene. Almost unnoticeable. But multiply it by billions of similar moments a day – and you get a gradual but fundamental shift in how people spend time, build connections, and endure discomfort.
The ability to cope with uncertainty – without immediately filling the silence – is no small thing. It's one of the basic skills of thinking. It's in the pauses that ideas are born, decisions are re-evaluated, and something resembling an inner life occurs. When every pause is filled by a system optimized for engagement, the space for this shrinks.
We aren't losing our minds. We are outsourcing them
The word «outsourcing» is usually applied to business. But it is the best description for what is happening to human cognitive functions in the age of smart systems.
Navigation. Spelling. Calculating a tip. Choosing a route. Picking words for a message. Assessing risk during a purchase. Deciding whether an article is worth reading – before even opening it. All of this is gradually shifting to systems that handle the task faster, more accurately, and without the slightest irritation.
It would seem to be progress. Liberation from routine. But there's a nuance that is easy to miss: many of these «routine» functions were not just tools. They were practice. The practice of orienting oneself in space – both physical and semantic. The practice of making small decisions, which builds the greater capacity to decide.
Neurobiologists have long known: the brain saves resources on what it doesn't use. A skill that isn't called upon atrophies. Not all at once. Unnoticeably. But irreversibly – if the break is long enough.
We aren't getting stupider. We are specializing in what systems can't yet do for us. The question is what will remain on that list thirty years from now.
Scene Two: The doctor who doesn't make mistakes
Somewhere in a major city clinic – it doesn't matter if it's Buenos Aires, Bangkok, or Mexico City – a system analyzes an image faster and more accurately than any specialist. This is a fact, and it is a good thing. Diagnostic algorithms are already detecting early stages of disease with a precision inaccessible to the human eye.
But here is what happens next. A young doctor, having grown professionally in the era of these systems, gradually stops training their own eye. Why bother – if the system is more accurate anyway? Years pass. The system updates. Then – for some reason: a glitch, a blackout, an outlier case – it becomes unavailable. And the doctor stands before the image alone.
This isn't a dystopia. It's the logic of any system of dependency. We've already gone through this with GPS and a sense of direction, with calculators and mental math, with search engines and memory. Each time we said, «It's fine, the tool just handles it better.» Each time it was true. And each time, we were capable of doing a little less ourselves.
Justice that no one verified
There is another conversation that happens less often than it should. It's about the fact that the systems we delegate decisions to are not neutral.
A credit scoring algorithm is trained on historical data. But historical data reflects historical inequality. If in the past certain neighborhoods were systematically denied loans – not because the people were less creditworthy, but because structural barriers were in place – the algorithm will learn that pattern. And it will reproduce it. Neatly. Without prejudice. Without malice. Simply because that's how the data is structured.
The same thing happens in hiring systems, medical diagnostics, and the distribution of social aid. The machine doesn't discriminate – it reproduces the discrimination baked into the data it was trained on. And it does so so convincingly that challenging the decision becomes almost impossible: «That's what the system said.»
When a human makes a decision – you can ask them. You can appeal to context, to an exception, to common sense. When an algorithm makes a decision – the appeal becomes a technical task, requiring access to code, data, and expertise that most people don't have.
This isn't the end of justice. It's its slow migration into a space that is increasingly hard to reach.
Scene Three: A digital twin at the wake
A few years ago, the first commercial services appeared that allowed one to «continue communicating» with a deceased loved one. A voice reconstructed from recordings. A texting style recreated from an archive of messages. Sometimes – synthesized video.
Technically, it is astonishing. Ethically – it is «terra incognita.»
Who decides when to stop? How does this affect grief – the natural, painful, but necessary process of accepting loss? What happens to the memory of a person when it is replaced by an optimized simulation that will never say anything inconvenient, never disappoint, and never die again?
We don't know. No one knows. The technology already exists and is in use – while the questions about its consequences are only beginning to be phrased.
This is the characteristic rhythm of our time: implement first, reflect later. Normalize first, ask questions later. By the time the question is asked, millions of people are already living inside the answer.
Concentration: Who holds the kill switch
There is a dimension to this conversation that especially rarely makes it into the public space. It concerns not what AI does to an individual, but what happens to power – in the broadest sense of the word.
The infrastructure on which the largest language models and decision-making systems run is concentrated in the hands of a few dozen companies. This isn't conspiracy – it's just economic fact. Developing systems of this scale requires computing power, data, and capital available only to a few.
This creates a new type of dependency. Not personal, but infrastructural. Hospitals, financial institutions, logistics chains, education systems – more and more sectors operate on platforms they do not own, by rules they did not set, with data they do not control.
When one of the major cloud providers experienced a massive outage a few years ago – hundreds of thousands of services worldwide stopped working for several hours. Airlines. Hospitals. Banks. Media. It was a technical incident, not a catastrophe. But it showed something important: the fragility of systems built on centralized dependency scales right along with the systems themselves.
Philosophers have long debated the nature of free will. But there is a more practical question that is now more important than abstract debates: how real are our choices if every one of them has been pre-filtered, ranked, and presented to us in «optimal» order?
When you open a food delivery app and see three options «for you» – you are choosing. But from what? From what the algorithm decided to show, based on your past behavior, the time of day, the weather outside, and hundreds of other signals. Your choice is real. But the choice-space was shaped before you ever opened it.
This isn't manipulation in the crude sense. It is – architecture. Invisible, yet all-pervasive. It doesn't forbid you from choosing differently. It just makes «differently» a little less convenient, a little less noticeable, a little less probable.
Multiply this across all spheres of life – the news you read, the people you meet on dating apps, the job openings you are shown, the prices only you see – and you get an environment where autonomy isn't canceled, but gradually eroded.
Scene Four: The classroom where the teacher is tired of explaining
Somewhere in a school, a child is writing an essay. It's a homework assignment. Twenty minutes later, it's done. The text is good – structured, error-free, with arguments. The teacher reads it. They feel something is off. Но they can't prove a thing.
This is no longer a hypothetical scenario. It's a description of what is happening in thousands of classrooms right now. And there is no clear villain here. The child didn't «cheat» in the old sense of the word – they used a tool. The tool works. The task is complete.
But what exactly was the task? To produce a text – or to learn to think? To formulate an argument – or to master the process of formulation itself? If the goal was the product – AI did a great job. If the goal was the process – the process was skipped.
Education has always been about productive effort. About the friction between what you know and what the task demands. It's in this friction that learning happens. When the friction is removed – a significant part of why the whole thing was started is removed too.
This doesn't mean AI has no place in education. It means the question of how it should be present there requires an answer – and it requires it now, not after an entire generation has passed through a system where this answer was never given.
What remains human
Everything said above is not an argument against technology. It is an argument against carelessness.
Fire is not to blame for conflagrations. A car is not to blame for accidents. A tool is defined by how and why it is used – and what control structures exist around it. With AI, the situation is the same, but the scale is different: never before has a tool penetrated so deeply into processes we considered fundamentally human.
Judgment. Compassion. Interpretation. Choice. Memory. Grief.
These things won't vanish – but they will change. They are already changing. The question isn't whether something human will survive in a world with AI. The question is how consciously we will participate in these changes – or whether we will simply allow them to happen to us while we stare at screens in waiting rooms.
When television appeared in the middle of the last century, critics predicted the end of reading, conversation, and the family evening. Some of those things did indeed diminish. Some adapted. Something new appeared in the place of the old.
AI is not the first tool to reformat the human experience. Но it is, perhaps, the first to do it so precisely, so personally, and so quickly that the adaptation of society cannot keep pace with the adaptation of technology.
It is here – in this gap between the speed of implementation and the speed of reflection – that the real risk lives. Not in a machine uprising. Not in the malice of corporations. But in thousands of small decisions, each of which seems reasonable, convenient, and harmless – but which, in total, add up to a world that no one consciously chose.
This is what danger looks like. Not as a catastrophe. As a habit.
It smells like morning coffee prepared in advance. It sounds like a notification arriving at exactly the right moment. It feels like ease – until you notice that ease has become the only criterion.
