← A Thousand Words
I had not actually intended to write about artificial intelligence as well. There are already more than enough people doing so with great competence and, at times, at considerable length. People who test every new model, comment on every announcement and know by Tuesday what the world will look like after Monday’s update. Then there are the sweeping promises, the warning voices, the astonishing images and the same recurring debates about whether this technology will save, replace or ruin us. One might therefore say that the field has been thoroughly cultivated. It does not need me as well.
And yet, here I am
The simple reason is that few subjects are so difficult to ignore. AI is no longer a specialist discipline taking place somewhere between research laboratories, data centres and science fiction. It writes texts, creates images, translates conversations, organises knowledge, programs software, shapes politics, threatens business models and appears in more and more of the tools we use every day. Some of this is impressive, some banal, some unsettling, and much of it still remarkably unreliable. But even after generously discounting the current hype, something relevant remains. Something that will change our working lives and probably our everyday lives as well. AI is already considerably more capable than Microsoft products or the iPhone were at a comparable stage of development.
No one knows how great this change will be. That may be the most honest statement one can make at the moment. We know the direction of individual developments, but not their destination. We can see how quickly certain capabilities improve, but not where limits will remain or new ones emerge. We can speculate about which professions will change, which activities will disappear and which new ones will arise. But speculation is not certainty. Anyone who confidently declares the ending of this story today is probably revealing more about their own hopes or fears. As ever, developments are lived forwards and understood backwards.
I do not want to turn either extreme into a programme
Neither cheers nor doomsday scenarios interest me very much. One side declares every new feature the beginning of a golden future in which machines relieve us of all effort. The other sees in those same features the end of work, creativity, truth or humanity itself. Both narratives have a peculiar attraction. They suggest clarity where none exists. They turn a confusing present into a simple story with an unambiguous ending. They are simply not very helpful in understanding what is actually happening.
I would rather take a closer look
I want to test what these systems can do, where they help and where they fail. I want to observe how my own actions and thinking change when a fast research assistant or astonishingly productive source of ideas is always available. I also want to know when convenience becomes dependence, when acceleration reduces quality and when a convincingly phrased answer creates only the illusion of knowledge. AI is not simply a neutral tool. The way we use it changes our expectations, our habits and perhaps even our idea of what constitutes a good idea, a good text or good work.
This cannot be discovered alone and in the abstract at a desk. One has to risk the experiment and use the things. One has to ask questions, assign tasks, examine results, look for boundaries and occasionally produce nonsense. That sounds obvious, but it is not. With a subject so charged in public debate, there is a strong temptation to acquire a finished position first and only then begin experimenting.
That is why I want to allow myself a certain naivety
Naivety here does not mean thoughtlessness. It is not an excuse to ignore risks or dismiss justified criticism. Questions of copyright, energy use, concentrated power, privacy, manipulation and the future of science and work are real. They deserve attention and challenge. But critical vigilance must be accompanied by a willingness to touch something in order to understand it fully: an openness that does not immediately force every experience into the framework of an existing conviction.
It also means being allowed to be wrong
Perhaps I will overestimate applications that nobody uses a year from now. Perhaps I will consider capabilities limited that become routine only months later. Perhaps I will discover a supposedly brilliant workflow and realise much later that I was faster without it. And I will probably publish texts whose judgements age badly. That is not an especially pleasant thought. The internet famously forgets very little. But an honest engagement with an open development cannot depend on always being right.
This blog is therefore not an attempt to explain AI once and for all. It is intended more as a public notebook for observations, experiments, mistakes and provisional conclusions. I am interested not only in what is technically possible, but in what these possibilities do to us and to me. How does creativity change when the first draft is effectively free? What happens to learning when answers are always available? Which work becomes more valuable when production becomes faster? And how do I retain judgement in a world where convincing results are easier to produce than reliable ones?
I want to remain optimistic. Not because a good ending is guaranteed, but because optimism creates room for action. It allows us to see technology not merely as something that happens to us, but as something whose use we can help shape. Openness does not mean approving everything. It means taking possibilities seriously without denying the side effects. Scepticism, in turn, need not mean standing still. It can be a form of attention.
Perhaps that is my real intention: to remain attentive while something is moving.
AI will not wait for an invitation before changing our everyday lives. It is already part of products, decisions and processes, often visible and sometimes barely noticeable. I do not want to comment on this change from a safe distance, nor surrender to it enthusiastically. I want to look inside, experiment, ask questions and describe what I find. As concretely as possible, as critically as necessary and with enough openness to leave room for surprise. And throughout this blog, I want to work directly with OpenAI Codex as a co-author.
I do not know whether this will ultimately produce wise insights. What is certain is that choosing not to engage would also be a position. Given the importance of this moment, it no longer feels appropriate to me. So I begin: without a finished map, without grand prophecy and with explicit permission to be wrong.