“If you have any work for me, please let me know.” This is how conversations now begin with young IT colleagues in Pakistan and India with whom I worked on WordPress and SEO only a few years ago. Back then they were completely booked out; I was more likely to be the one asking for help whenever I needed support on a project. Today, things look very different there. AI already seems to have taken away their work across a broad front.
As I understand it, the women there are trying to do things with AI themselves. Yet the reshoring of their previous work appears to have happened long ago—not physically, but in the digital realm. The work has not moved to another low-wage country. It may simply be performed where the client is based, using machines the client can now operate directly. Pakistan and India as workshops for websites, SEO and related activities seem, at any rate, to be dissolving.
That is remarkable precisely because of the speed of this disruption: we are talking about two, perhaps three years, not an entire generation. It is a signpost for what may soon be expected elsewhere, including here in the EU.
Anything is possible
Anyone working with modern commercial AI systems today—as I have been doing and implementing for some time—already knows their potential to replace much of the work done by cost accountants, estimators, dispatchers or payroll clerks, or to reduce the human effort required to a fraction of what it was.
People are often misled by the ChatGPT or cloud interface, where all they see is a dialogue, and say: well, that cannot really do anything. In doing so, we underestimate what is already possible behind the scenes. What can be done. What is being done. Interfaces to a wide range of systems; connections to telephone systems through which AI can actively call people or request information; email systems in which it reads and answers messages; or browser control that lets it access other database systems even when no direct interface exists.
In other words, we already have a system that could, in principle, join a Teams meeting as a speaking and listening colleague—scheduling appointments, completing tasks, requesting information, writing emails and calling people. Technically, this is where we are today.
People will say: yes, but what about the human spirit, the ability to solve individual problems? In interpersonal relations, probably. Yet in compliance questions or the search for possible solutions, the machine is already better in many fields because it has access to a larger pool of information and can connect possibilities that are not readily available to any single person.
So why is nothing happening yet?
If we now consider that the European Union has, incredibly, around 125 million white-collar jobs—jobs performed by people who are essentially occupied with information, organisation and decisions—and that a significant share of these activities may be at risk, the first question naturally becomes: why has it not happened yet?
Why are we not experiencing the great collapse right now, at this very moment? Why have people not long since lost their jobs? Why are people still sitting in tax advisory firms, still working as cost accountants in companies or as dispatchers assigning workers?
There is always a certain inertia that resists anything new or any form of change, especially when that change threatens the existing system. To me, this seems like an inertia inherent in systems—almost a natural law that makes them want to continue in their existing form. You know your processes, your staff and your risks, and as long as things work somehow, you carry on. The system rarely rewards change or an impulse from outside.
This inertia is overcome when the pressure becomes great enough. And it is precisely here that the road divides: disruption, meaning radical acceleration, and erosion, in which the existing system is hollowed out step by step. Neither version is especially pleasant or attractive, but distinguishing between them may help us understand what is now approaching.
Two speeds
What might create that pressure? Imagine ten tax advisers. One begins to eliminate four of five payroll positions, automates most of the work and can offer a payroll calculation for three euros instead of twenty. The other nine cannot compete with their existing staff because their labour costs are too high. They therefore have to find similar AI solutions—not necessarily because they are enthusiastic about technology, but because the market leaves them no choice.
That would be the disruptive scenario. What is fascinating is that prices fall as well. Cost reductions are passed along entire service chains, margins come under pressure, and AI can also create an oversupply of certain services, pushing prices down again. We may then face not only a loss of jobs but a real contraction of revenues and economic structures, perhaps even a broader shrinking of the economy.
That makes the next question difficult: how do we support people who lose their jobs when the economic base on which additional taxes would be levied is itself under pressure? People often talk about robot taxes or other new levies. But imposing taxes on a shrinking economy does not automatically produce the additional revenue that would be needed.
Quite apart from the fact that Europe’s maximalist states—and here I mean the EU—are already extremely overbalanced and control large parts of the economy. On top of that, in some fields they have long been pushed to the margins of globalisation. This may yet matter in the end, and perhaps in a rather unpleasant way.
The second possibility is erosion. Businesses begin to experiment, use AI where it is technically easy and works with reasonable reliability, and do not replace employees overnight. Instead, positions that become vacant through retirement or resignation are simply left unfilled. People hesitate to jump into cold water because something might not work, a process might come to a halt, responsibility remains with them, and no one knows who will have to answer for the error.
Ultimately, the result can be the same, only slower. The contraction may be less severe at first, leaving more time to respond—or, more precisely, to soften the impact and search for alternatives. Yet in either scenario, the disappearance of certain jobs seems almost impossible to prevent.
And what happens to people?
What happens to people depends again on whether we experience disruption or erosion. Rapid disruption could mean a plunge into unemployment and a society in which many people receive a basic allowance while doing small everyday jobs on the side: gig work in which they collect or obtain something for someone, perform whatever the AI or drone cannot yet do, and receive a few euros in return.
It would be a highly atomised, barely hierarchical gig economy of a kind we also know, in another form, from the informal economies of South America or Africa. If unemployment were to rise to 40 or 50 per cent in such a scenario, this is exactly what might emerge. My colleagues in Indore or Multan already seem to have slipped from longer-term project work back into individual gigs of this kind. Perhaps we are seeing there, on a small scale, what still lies ahead elsewhere.
If the change unfolds as erosion, by contrast, there is at least hope that its slower pace will allow other, parallel developments to emerge and absorb these people or groups of people. I would not look first to state planning, because it generally performs rather poorly in changes of this kind. But there may be forms of employment, business models and economic activity that we do not yet know and therefore cannot foresee.
Only one thing will not work: employing people on a large scale in state-funded entertainment jobs and believing that this has solved the problem.
What should I do?
What can we do for ourselves and for those around us? I believe we should first observe very carefully whether we are moving more quickly towards disruption or towards erosion. And we should examine which jobs might survive only in rudimentary form and which could disappear altogether.
There are probably countries that offer a certain model—or, rather, serve as signposts. They are likely to be the United States and several Asian countries. Anyone looking at China or different US states today may recognise what is happening sooner, and we must align our economic lives accordingly, perhaps with a radicalism we have not previously known in our European lives. Our comfortable European lives.
And what exactly does this mean for an entrepreneur? It means looking for leverage in one’s own company and one’s own job. Gathering information, working with people who have already implemented something, and learning to recognise which parts of the systems in which we work are suitable for AI. Not selecting one small task, but looking at the system as a whole.
This requires something that, amusingly, many AI sceptics lack: creativity and imagination. The more we can imagine automating, and the more boldly we approach it, the better we will do it. And perhaps we will then have more time for the things that truly give us pleasure. There is no question about that either.
What matters, however, is that courage, that creativity and the ambition to think big. Someone who says today, “Well, for now I only want to automate this one letter,” might just as well say, “Fine, I want to automate all my letters. I also want every letter that arrives to be scanned, analysed and answered.” Or I go one step further and say: “Hey, read my emails, analyse them, assign them to colleagues and obtain feedback after three days.” Then we are no longer talking about an automated letter. We are talking about a workflow that largely runs itself.
Generation AI
Do not close yourself off to this development. You cannot sit it out. It is coming. Do not surround yourself exclusively with people who explain why something will not work. Do not end up on the losing side. Think clearly and analytically, approach the matter aggressively and, above all, be willing to question yourself and your own processes. Perhaps be able to imagine a new life as well.
And, of course, speak to people who are already implementing such systems—including me and my consultancy, Generation AI. Because what is coming now is Generation AI, and you should be part of it.
These are not things we can truly choose. But we need to talk about them as soon as possible.