The problem is not that robots have become like people.
The problem is that people have started living like robots.
Open up. The bear has arrived
It seems impossible to open a news feed today without seeing headlines such as “AI replaced thousands of employees,” “these professions are no longer needed,” or “ChatGPT has learned to do this or that.” The headlines are usually exaggerated, but behind them there is still a real and unsettling trend.
According to forecasts cited by McKinsey, by 2030 up to 30% of office roles may be partly or fully automated. In the United States, around 14% of employers already use AI for tasks previously performed by people. In Japan and South Korea, the figure has moved beyond 20%. I suspect the share is even higher in content production, technical support, and online consulting. Russia even has a national AI development strategy through 2030.
When ChatGPT 3.5 became the first truly mass-market AI product in 2022, it triggered a new wave of digital transformation in the workplace. More capable models followed, and the direction became obvious: AI in every home, every profession, and every process where it can be applied.
No one appears ready to stop. Analysts expect trillions of dollars to be invested in AI development before 2030.
This is not another story about a machine uprising—although, just in case, my apologies to Skynet. I want to understand what place may remain for people in that future.
Forewarned and unarmed
Back in 2003, long before GPT models, Tom Peters published Re-imagine! The book explored many ideas and is still worth reading today, but it paid particular attention to the automation of work. While most people were watching blue-collar jobs, Peters looked a little further:
Blue-collar workers are already on the way out. White-collar workers are next.
The age of mediocrity, repetition, and templates is ending.
He described ordinary employees who spent years moving bolts or documents from one place to another without asking why. They were the first to be replaced. After all, if a job can be fully described in an instruction, it can probably be automated.
Do we confuse flexibility—the absence of instructions—with fear of losing control through automation?
At the time, many of Peters’s statements sounded like slogans or provocation. I did not fully understand the scale of his argument either. A line such as “Doctor, this one-dollar chip can already read an ECG better than you—what will you do about it?” was striking, but it still felt more theatrical than realistic. His concept of “I am the office”—working from anywhere in the world—mostly inspired mild envy in 2003.
I understood the warning and felt the anxiety behind it, but I could not yet see the whole picture. The depth of the coming changes was hard to sense. The threat felt too abstract, especially in IT.
Peters’s main point was not that people are valuable because they can follow a process. Their value lies in being able to change the process. Initiative matters more than instruction. Thinking matters more than compliance.
Routine expertise? Not interesting.
To understand what is happening to work now, it helps to look back. We have already been through something similar during the first evolution of labor.
Digitalization says hello: the first evolution of labor
The twentieth century brought a vast transformation that permanently changed industry. Automation entered factories. Machines and mechanisms gradually displaced manual work that people had performed for decades. Workers feared that soulless pieces of metal would take their jobs, and those fears were not baseless. Individual occupations disappeared, along with entire industries built on repetitive operations. Everything that could be standardized was replaced—when the economics made sense.
Yet the disappearance of old roles created new ones that were more complex and required different skills. Industry needed people who could operate, maintain, and configure machinery. A blacksmith once forged a part by hand; now a stamping press forms it under an engineer’s control. A telephone exchange once relied on operators connecting wires; now it requires specialists in digital communication systems.
These jobs did more than replace earlier ones. They raised expectations for thinking, knowledge, and adaptability. Work began moving from physical effort toward intellectual effort.
Not everyone managed to adapt, and some people were left behind. At the same time, the transformation accelerated education, retraining programs, and applied science. Craft production declined when it could not compete with assembly lines, but demand grew for engineers, programmers, systems analysts, and teachers capable of preparing a new generation of specialists. Human resources evolved from routine personnel administration toward talent management.
Automation destroyed thousands of jobs, but it also created room for hundreds of thousands—or millions—of new ones. Paradoxically, I have often seen automation increase headcount. My explanation is that higher productivity accelerates the entire business, and the company then needs more people to support and develop operations at the new level.
The main condition for survival became the ability to change, learn, adapt, and transform again and again. The motto of the first wave was simple: learn throughout your life.
This painful but necessary stage of labor evolution triggered fierce resistance. Think of the Luddites in nineteenth-century England, who destroyed machines they blamed for their hardship. Decades later, offices repeated the same pattern—from sabotaging electronic document systems to fearing digital assistants.
The irony is that people who moved from factory floors to computers often became the new blue-collar workers, and their work was immediately targeted for automation.
The bet on physical effort—including stubborn effort performed at a computer—was lost. People shifted toward intellectual work. Knowledge, reasoning, and analysis looked like a safe new advantage. Surely machines could not compete there. But progress went in exactly that direction, and it no longer mattered whether a computer sat on your desk.
The second wave: knowledge is no longer power
We have reached the point where Peters’s prediction is becoming literal because of AI. One person with access to the right tools can now do work that once required an army—from building a website to preparing a strategic plan. This is no longer an abstraction. Many business problems are becoming questions of tooling rather than company size.
AI can already perform tasks that were recently treated as strongholds of intellectual work:
- drafting legal documents;
- analyzing large datasets;
- generating emails, articles, and social posts;
- consulting customers;
- consolidating reports and assembling presentations;
- designing and generating images, video, and voice;
- writing and testing code.
Yes, AI fails, sometimes badly. Its answers contain errors. But each generation improves. The output that made an entire office laugh today may soon make the same people return quietly to their desks, slightly unsettled.
Something tells me that within the next few years, the quality of AI work will match the output of an average specialist who still needs an occasional push to deliver an acceptable result.
Only AI revealed how much intellectual work is also repetitive and routine.
AI is at home wherever thinking has been reduced to repetition rather than exploration. It enters the places where a person has already turned into a function.
If you were once valued for answering quickly, you will now be valued for going deeper. Answers are no longer scarce. Reflection is replacing erudition; meaning is replacing speed.
Many problems are now questions of tools, not the number of people in a company.
We are losing the second wave of digital labor transformation with the old bet: “I know, I can, I am capable.”
An attack on creativity: the third knockout
The first wave struck physical work. The second targeted intellectual work. The third, paradoxically, is moving toward creativity—the territory we still treat as humanity’s final refuge. These are precisely the capabilities valued today: asking questions, producing ideas, defining strategy, and reframing problems.
Does that still sound like science fiction?
Consider the growing popularity of prompts such as: “Ask me everything you need to give the best possible answer.” Users report that this genuinely improves the result. Some AI systems already end an answer by asking whether they should add, improve, or reformulate something.
People increasingly want AI to initiate the dialogue, understand the task, and guide the process—in other words, to ask the questions itself. AI is moving from reaction toward initiative. While I was writing the original article, an AI agent was released that could act as an assistant for email and calendar management.
You still have to ask the agent to act, so it remains reactive. But how long will it take before we hand over email, messaging, and phone management completely? That would be another small step toward a personal AI.
A scenario that recently seemed impossible is becoming realistic:
AI analyzes a company’s data and notices that sales of a product are falling. It proposes improving the funnel, creates a simple landing page, writes the headlines, runs A/B tests on images, optimizes SEO, launches the site, and watches the statistics. It then adjusts the copy, changes colors, and sends offers to leads through the channels it has identified as most effective.
Along the way, it may assign tasks to employees and monitor completion.
Who would trust a machine with that? Dell, for example, was already using supply-chain systems in the 2000s that could automatically negotiate discounts on other products when an order was seriously delayed.
At Netflix, AI influences not only which shows are made but which scenes are selected for trailers based on the preferences of millions of viewers. At Amazon, AI makes hundreds of decisions every second—from pricing and warehouse placement to storefront personalization and logistics capacity. At Alibaba, AI can distribute advertising budgets in real time, choosing channels, creative assets, and targeting, then reallocating money when a campaign underperforms.
It will be amusing when another AI appears on the opposite side of the transaction. Will the two systems negotiate a discount, or simply close the deal in silence using rational criteria—faster than any sales department could after dozens of strategy meetings and business trips?
Broad adoption is still some distance away. The main struggle is still happening at the second boundary, but the outline of what comes next is already visible.
When AI becomes creative enough to replace a marketer, screenwriter, designer, and analyst at the same time, the third boundary will fall. Creativity will no longer be an exclusively human privilege. Some organizations have already experimented with presenting AI as a CEO, including NetDragon Websoft and Dictador Holdings.
What remains for us?
Even when AI begins formulating its own questions and initiating tasks, I believe it will still need context and meaning. Meaning differs between companies just as motivation differs between people. AI will need to understand what you actually want—the answer to the question “why?”
The third wave may therefore produce a kind of meta-human: someone who defines the metadata, constraints, culture, and purpose within which AI operates. Industry-specific AI will still need a person to define the industry and the reason for acting.
The fourth wave: creating the creator
The future does not cancel people. It changes what people are paid for.
I am not sure we will live to see the end of the third stage, but it is still interesting to ask what comes next when that bet is lost as well.
Imagine a meta-human who defines the framework, context, culture, and purpose for AI. The system then acts autonomously to implement that context.
Suppose someone says, “I want to make money in freight transportation.” A few days later, the person may only need to sign an application for a digital signature and then watch the company develop—like a complicated strategy game—while resisting the urge to intervene in operational and strategic management.
Some businesses fail not because their owners pay too little attention, but because they pay too much.
In the fourth wave, AI would have to understand what is beneficial to an individual, a state, or humanity as a whole. It would create the context and then proceed as before. This resembles a digital hierarchy: person, community, institution, civilization.
At that point, the remaining human role may be to create virtual civilizations—complex models of possible worlds. Whether people will still be present in the conventional physical world is another open question. Perhaps this is where digital and human consciousness begin to merge.
I would be interested to hear other hypotheses about the third and fourth waves.
Or perhaps some people, overwhelmed by digital singularity, will reject it all, move into forests and fields, form tribes, and grow potatoes in the old reliable way. Centuries later, archaeologists may find the remains of their structures and wonder why an advanced ancient civilization built them. A technology can be forgotten completely in roughly three generations.
People once flew inside giant birds that drank special water and breathed fire from their wings.
Dad, I love your stories.
Where is the proof?
An attentive reader will notice that everything described here as the past or the future already exists somewhere in the present.
Some people still rely entirely on manual labor. The people of North Sentinel Island remain isolated from modern civilization and continue living with bows and arrows. One-person companies exist. So do trendsetters, subcultures, and people who leave society to build a new life from scratch. All of this happens at the same time on one small planet: people run with spears and fly into space.
The future arrives unevenly.
It comes in waves—not everywhere, not simultaneously, and not for everyone. In one place we see yesterday; in another, tomorrow. Sometimes both fit inside the same room: an iPhone in one hand and an abacus on the table. The world is archaic and futuristic at once. We only need to look closely.
We often mistake the new for a miracle, although it is usually another version of something old. Nature demonstrated zipper-like mechanisms in mollusks long before we copied them. Rockets and aircraft are new kinds of horses that move us from point A to point B. AI and robots are not entirely new either.
Instead of conclusions
Show me your browser history, and I will tell you who you are.
I want to end where I began. AI is neither an enemy nor a friend. It is our projection—an image of ourselves, our culture, and our knowledge. AI is our reflection in the mirror of the information space. Our patterns, mistakes, and successes are built into the foundation of how it works.
There is no need to fear our reflection. This is another evolution of labor:
- Follow orders. Physical labor, instructions, and assembly lines—defeated by mechanization and automation.
- Answer questions. Analysis, reports, and decisions—currently being displaced by AI.
- Ask questions. Strategy, reframing, and creativity—likely to be challenged by AI capable of proactive action.
- Create context. Culture, boundaries, metadata, and systems of values—eventually challenged as well.
New industries tend to travel through a similar sequence, whether agriculture, mechanical engineering, or IT. They begin with repetitive labor; automation arrives; the process continues.
Can it be stopped? I do not think so. Remember how painful the transition from gas lighting to electricity was. Our streets are now illuminated electrically, and even traditional bulbs have largely given way to LEDs. Progress can be slowed—sometimes significantly—but that is probably the limit.