The development of artificial intelligence over the last several years has caused something that is difficult to call anything other than an earthquake—an earthquake encompassing vast areas of professional work: machine learning, data analysis, accounting, document handling, translation, and that entire layer of work which, fundamentally, consists of transferring information from one format to another. The point is that this earthquake is visible in employment statistics and headlines, but not where it should be most visible: in the everyday work of ordinary companies, workshops, wholesalers, offices, and accounting firms. The tools lie there waiting. Few people pick them up—very few.
This is the real subject of this article. Not what artificial intelligence can do, because more has already been written about that, but why we, despite having it within reach, behave as though it did not exist.
Innovation Is Born at an Intersection, While a Person Usually Knows One Field—or Two at Most
It is difficult to be an innovator in every field simultaneously because it is difficult to know every field simultaneously. This is a truism, but more follows from it than may appear. Innovations very rarely arise within a single specialization. They emerge at an intersection: someone who knows the need meets someone who knows the tool; or—something that happens less often and produces spectacular results—one mind contains both the need and the tool. An engineer who understands code. An accountant who understands statistics. A doctor who understands how a database works. There are few such people; we educate them accidentally rather than systematically, and organizations can waste them within six months.
Here is something worth stating directly: a language model is trained on practically all fields at once. It does not know them as well as a specialist, but it knows all of them well enough to see a bridge between disciplines that no individual person holds in their mind at the same time. From the perspective of creating innovation, this is precisely the capability that we humans lack. It is easy for a machine to combine logistics with queueing theory and invoicing with image recognition, because to it these are not distant worlds, but adjacent fragments of the same text.
Why, Then, Does the Machine Invent Nothing by Itself?
The answer is disappointingly simple: because it has no reason to. Artificial intelligence is passive. It waits. It sits and waits for our question, our idea, our suggestion, our command. It has no interest of its own, no anxiety, no sense that something should work better, because nothing bothers it. It does not wake in the morning thinking that three people in this company copy data from a PDF into Excel and that this is absurd.
Perhaps that is fortunate. If it had its own motivation to rebuild the world, no one knows how that would end, and I suspect that no honest person can predict it. Since it is passive, however, one might expect us to be active. This is where the real problem begins, because we are not active—and we have very good, very old reasons for that.
A Million Years of Training in Conserving Energy
Evolution encoded survival within us, and survival largely comes down to managing energy. Do not expend more than necessary. Do not take risks without a reason. Do not run if you can walk. This program still operates within us—in the office, on the production floor, at a management-board meeting—and it cannot be switched off by a declaration.
The paradox is that automation is exactly the same thing: saving energy. The purest automation in the history of hunting was setting a trap. The animal falls into it while the hunter sleeps. A fishing net catches fish without a person. Sowing is also automation, and one of the greatest forms we ever introduced: grain grows where someone placed it, so there is no need to search for it throughout the area. The entire Industrial Revolution was nothing more than a series of cycles in which another automation appeared, cut costs, increased profit, and improved the capacity to survive—of companies, families, and individuals.
In theory, then, evolution should have created us specifically to introduce innovation. Yet innovations do not arrive quickly or easily, and the question is where precisely the barrier stands. It does not stand in technology. It stands in three other places.
Barrier One: The Environment, or the Employee Who Prefers Silence
People dislike change, while organizations, contrary to what they say about themselves, do everything to prevent change. Grand stories from HR departments about collecting employees’ ideas, rewarding initiative, and counting every voice are worth roughly as much as declarations about treating women and men equally, or assurances that employees of every age—especially very experienced ones—are hired with enormous enthusiasm. Everyone knows what this looks like in reality.
The mechanism is deeper than ordinary hypocrisy, however. Proposing an innovation exposes a person to personal, concrete, measurable risk. If the idea fails, it will be that person’s idea. If it succeeds but deprives someone of part of their responsibilities, the innovator will make an enemy. If it automates the work of three people in the next room, the innovator will have to look them in the eye every day. The first rule of our internal program is to avoid risk. This is why an employee who has three solutions to difficult problems in their head usually keeps them there. Not because the employee is lazy or stupid, but because the employee is rational.
Barrier Two: Organizational Risk, or Who Will Build the First Mill?
Imagine a settlement two thousand years ago where grain is ground using quern stones. Someone comes up with the idea of connecting a stone to a waterwheel. A wonderful idea. The trouble is that a mill wheel requires hundreds of hours of work: trees must be felled, timber shaped, a structure built, gearing made, and a dam constructed on the river. It is an effort requiring a group of people for several weeks—effort taken away from other activities in a world where no one has spare time.
Someone had to build it first, with no guarantee, because nothing like it had existed before. What if the structure did not turn? What if the dam burst? Even today, I find it difficult to imagine the moment when someone first believed that crude wooden gears would rotate because equally crude blades would catch the resistance of the water. This is precisely why humanity took hundreds of years to create even these most basic inventions. There was no shortage of ideas. What was missing was someone willing to bear the cost of the first attempt.
Nothing has changed in companies in this respect. Only the scale of the cost has changed: today, the first attempt costs a dozen or so hours and several hundred zlotys rather than several weeks of work by an entire village. The reflex remains the same.
Barrier Three, the Worst One: The Horizon
For approximately three years, I have written that we can have our own AI assistants—assistants that manage the calendar, reply to emails, arrange the day’s schedule, keep track of matters, and bring order to our lives. Three years ago this was difficult because it required programming, building separate applications, and several dozen hours of work by people with appropriate qualifications. Today it is a matter of several hours of configuration, largely free of charge and without writing code. Almost no one does it. A few people do.
They do not do it because such a change does not add itself to their existing life; it rebuilds its structure. One must plan the day differently, make decisions differently, distribute tasks differently, and think differently about what constitutes one’s work and what does not. It disrupts the horizon, and people’s horizon is astonishingly heavy, as though made of lead.
Videoconferencing provides the best example. Technology capable of supporting meaningful remote meetings had existed since at least 2005. The tools were immediately available, ready to use, and inexpensive. Yet businesses continued to travel to customers, meetings were held in conference rooms, and suggesting a conversation through a screen was regarded as disrespectful. Only the pandemic—a catastrophe, compulsion, a situation with no alternative—made this form obvious within several weeks, and today it is indispensable. Fifteen years of waiting for change, and those several weeks of the pandemic. The technology did not change then. The horizon changed.
Catastrophe as the Only Proven Accelerator
History is unpleasantly consistent here. The Second World War produced jet engines, rocket propulsion, radar, and nuclear weapons. The space race and the Cold War added materials, plastics, computer technologies, and data transmission—the foundations of everything we use today. In less than three years, the war in Ukraine developed a new form of warfare in which inexpensive drones controlled by increasingly autonomous software proved more effective than expensive equipment designed for the previous era.
The conclusion is bitter: comfortable living does not motivate technological development. Development accelerates when defeat is the alternative. This means that we wait for our own catastrophe as a prerequisite for progress, which is a strategy as effective as it is idiotic.
What People Say and What They Do Not Do
For at least two years, at conferences, during panel discussions, and over coffee, I have heard the same thing: we would like to introduce artificial intelligence. In the warehouse. In logistics. In supervision. In production. Incidentally, what they discuss is usually slightly outdated and has certainly been implementable for several years, but they speak this way because this is how they understand it, and it is difficult to blame them. Yet the readiness is absent. Entrepreneurs are tired, they are not technologically oriented, and a day contains only as many hours as it contains.
Then the most convenient idea of all appears: perhaps artificial intelligence should be forced to become active. Let it find opportunities itself, propose improvements itself, and implement itself. Someone should build such a system. Specialists should do it. We need only find them, persuade them, explain matters to them, finance them, wait until they do it for us, and then have the finished artificial intelligence devise and implement everything for us—while we, with our leaden horizons, either finally move or become frightened and sweep the entire matter under the carpet.
No. That is not the point. It is simply one more excuse, only in newer packaging.
What Can Be Done Today, Without Specialists and Without a Budget?
The newest models—from the Anthropic, OpenAI, and Perplexity families—are good enough that one need not work with them through a form containing fields. One can conduct a long dialogue with them. One can tell them the entire history of one’s business, all its problems, and all the places where something becomes stuck, just as one would tell a spouse or close colleague. The model will record it, organize it, and not forget it.
One can then ask for five points where something can be changed, simplified, or automated. Those five points will be provided. This is the breakthrough moment, because with a concrete list one no longer stands before a fog called “we ought to implement AI,” but before five tasks. Implementing them is no longer difficult either: building a simple application, integration, or automation takes hours rather than months and costs relatively little. It is enough to begin with one point, the smallest one, to discover that nothing bad happened.
Thus neither the lack of specialists, fear of losing work, organizational risk, nor our evolutionary energy-saving program—none of the things that successfully blocked our development for millennia—justifies passivity today. We must rid ourselves of narrow horizons and stop thinking according to templates. Artificial intelligence is better suited to precisely this purpose than to almost anything else: as an interlocutor with no interest in preserving the status quo.
Finally, a question that usually gets lost in texts of this kind: why do we need this development at all? So that we do not have to work so hard. So that we have time for our families. So that we earn more with less effort. And so that we can decide about our own lives to a greater degree than when we are overworked, constantly busy, and squeezed into conditions that none of us consciously chose.
Wojciech Moszczyński
Wojciech Moszczyński—a graduate of the Department of Econometrics and Statistics at Nicolaus Copernicus University in Toruń, a specialist in econometrics, finance, data science, and management accounting. He specializes in optimizing production and logistics processes. He conducts research into the development and application of artificial intelligence. For years, he has been involved in popularizing machine learning and data science in business communities.