
People like to talk about AI “hallucinating.” It sounds clever, but it is sloppy language. Hallucination belongs to the human mind. What we are dealing with here is a system that produces plausible nonsense when it runs out of solid ground.
Another familiar complaint is that AI generated images look “generic.” That tends to say more about the prompt than about the system. A generic instruction will reliably produce a generic result. There is nothing mysterious about that.
AI is built to be helpful, but it does not operate in a vacuum. It needs direction. The quality of that direction largely determines the quality of the outcome. Model size and architecture matter as well, but with systems such as Claude and OpenAI you are already looking at the upper range of what is currently achievable in terms of usefulness. I was once impressed by Manus.ai during a specific research task, but that is a side note.
Helpfulness is the core, and people translate that into a familiar fear: AI will take over my job. What is harder to picture is that the work itself changes shape. People rarely enjoy that idea. As a photographer, I did not greet the smartphone with enthusiasm either. Within a few years, large parts of the lower and middle segments of the market collapsed. In a small country like the Netherlands, that effectively hollowed out the profession. What remains visible at the top are a handful of fine art photographers and highly specialized niches, while the broad middle ground that once sustained many careers has largely disappeared.
I did not put my cameras away. I changed the way I worked. That process was slow and uncomfortable, especially at an age where flexibility does not come naturally anymore. Today I run one of the last small photography studios in the center of Amsterdam that is still open. You cannot walk in, but by appointment it works just fine. I reorganized my practice, partly by publishing my own work. In practical terms, I now approach the customer instead of waiting for the customer to approach me. The income is not what it used to be, but the pleasure in the work survived, and that matters more than people tend to admit. It feeds both mental and physical resilience.
AI is, first and foremost, a tool. A sophisticated one, but still a tool. And a tool is useless in untrained hands. You may prefer traditional craft and feel no urge to embrace these developments. That is your choice. It simply is not a neutral one. It comes with consequences. Progress may be an overused word, but the drive behind it is unmistakably human.
How this plays out is your responsibility, and largely a question of inventiveness.
I often think back to a Japanese CEO of Canon who introduced a new motto every year. The year I attended his New Year speech, the message was simple: be modest and be flexible. It still applies.
Of course money flows will shift. They always do when technology moves. Not everyone will benefit. That part is not caused by AI itself. It is shaped by policy, by markets, and by political choices. Blaming the tool is convenient. It is also inaccurate.