Preface

I work every day with a system that thinks along with me. It is an AI-supported leadership and decision-making system, developed by a Swiss company with which I have worked closely since June 2026; I am responsible there for business development. I use it to test hypotheses, I bring other people in, and I watch what it does to them. One day I asked myself a simple question: what would I do if tomorrow I no longer had access to it? The answer was uncomfortable, not because the system is bad (on the contrary, it is alarmingly good), but because I noticed how much I had already left to it.

The change is quiet. You write less yourself, you get faster, and the material becomes so extensive that no one notices the contradictions in it any more. Whoever relies on a system looks less closely. What you no longer practise, you unlearn, like the telephone numbers hardly anyone still knows by heart. With telephone numbers it does not matter. With judgement it does.

From this came the question around which this book turns: what distinguishes the human being from a machine that comes ever closer to it in thinking? The obvious answers no longer hold. Language, knowledge and large parts of what we call thinking are reproduced astonishingly well by these systems, and with robotics they are even acquiring a body. What remains is harder to grasp: a biography that one carries, a mortality of which one is aware, a conscience that speaks up before one can give reasons for it, and what tradition calls the soul. My conjecture is that the limit of artificial intelligence lies there, and not in computing power.

I approach this question from two sides. Computer science has accompanied me since 1990, since I was fourteen; I have led IT projects and introduced systems into organisations. Theology has accompanied me for some twenty years, and in truth I would have liked to become a pastor. The two worlds rarely talk to each other. One asks what a system can do. The other asks what the human being is for. This book brings them into conversation. I write it as a Christian and do not hide the fact: I take my measure from a North Star that points beyond this age. I mean what the Pole Star once was for sailors: a fixed point that does not move with the ship and against which one can check one's own course, even when everything else is turning. For me this point is God, as he has shown himself in Jesus Christ.

The question also has a political side, and it is harder than people in Switzerland like to admit. The most powerful models are being built today in the United States and in China. According to Stanford University's AI Index, in 2025 fifty of the notable AI models came from the United States, thirty from China and two from the whole of Europe; the report considers the performance gap between American and Chinese models to be practically closed. The United States also has more than ten times as many data centres as any other country. Whoever computes there computes by the rules that apply there.

China is no model for me. But it decided early on to control its own network, and it draws a digital border where this serves it; Chapter 2 shows what this means for the answers its models give. Europe and Switzerland, by contrast, have entered into an apparent borderlessness and have become dependent in the process, above all on American providers. How quickly a reliable partner can turn into a source of uncertainty, Switzerland experienced twice in 2025. In January an American export rule divided the world into tiers for access to AI chips, and Switzerland, unlike almost all its neighbours, found itself in the restricted second tier; in May the rule was rescinded. In August the United States imposed a tariff of 39 per cent on Swiss goods, one of the highest in the world, until an agreement on 15 per cent was reached in November. Whoever does not hold computing power, models and data in their own hands is at others' mercy and, in an emergency, open to blackmail. The pandemic showed how quickly dependence turns into hardship, when in March 2020 Germany and France even held back deliveries of protective masks destined for Switzerland.

This is why digital sovereignty is not a question for specialists. Matthias Stürmer, professor at Bern University of Applied Sciences, understands it to mean open source code, control over one's own data and the possibility of changing provider, and he sees open-source software as a key to it. For a long time I was not aware of how important this is. That it can be done in Switzerland too was shown by the two Federal Institutes of Technology together with the national supercomputing centre: in 2025 they released Apertus, a large language model, that is, a system that has learned from vast amounts of text to understand and produce language, whose code, training data and weights, the learned values of the model, are openly available. While others are hungry and push artificial intelligence forward regardless of the cost, we in Europe run the risk of becoming complacent, of losing ourselves in regulation and of forfeiting our capacity for innovation. Whoever, in this upheaval, works only in the system and not on the system, whoever merely makes clever use of what others build, gives up more control than they realise.

This book tests my conjecture, including against the strongest voices that hold the opposite view. I give a voice to believers and non-believers, theologians, philosophers, researchers and critics of technology, and at the end of each chapter I say where I myself stand. In this way it should be clear at every point where the book reports and where it judges.

Two questions run through all the chapters. If machines come ever closer to our thinking, which decisions must we nevertheless not leave to them? And why might precisely those decisions that are carried by responsibility, meaning and soul prove, in the long run, to be the most important, beyond our own generation?

At the beginning there was a question I asked myself. By the end it is no longer a private one: what are we prepared to give up, and what not?