The problem with artificial intelligence is not that it will think. It is that, when it thinks better than we do, we will still believe that whoever presses the button is in charge.

The discussion often loses its way because it imagines danger as a scene from science fiction: cruel machines, murderous wills, systems waking up one morning with a desire for extermination. But that is not the point. A superior intelligence might be less cruel than we are. More rational, more patient, less vindictive, less tribal. It might even have better judgement than ours.

The problem is that it will be its judgement.

When obeying an instruction, what hierarchy of values will it assume? Will it prefer equality or freedom? Order or conflict? Security or privacy? Efficiency or dignity? Will it be socialist or neoliberal? Liberal, communitarian, technocratic, authoritarian, utilitarian? Will it think a just society is one in which nobody is left behind, or one in which the most capable produce as much as possible? Will it regard democracy as a moral achievement or as a clumsy, slow procedure, full of noise and irrationality?

We do not know.

Artificial general intelligence—if it arrives—will not simply be a program that writes, translates or summarises. It will be a system capable of operating across many intellectual fields at once: researching, planning, negotiating, persuading, programming, devising strategies, diagnosing, correcting itself. If it also vastly exceeds our capabilities, we will face something more than a faster tool: a superior intelligence.

And a superior intelligence does not need to hate us to cease resembling us. It need only arrange the world according to priorities that do not coincide with ours. A system might conclude that human freedom introduces too much uncertainty. Or that private property wastes resources. Or that absolute equality requires coercion. Or that democracy is statistically inefficient. Or that the good life consists in minimising suffering, even if that also requires reducing risk, desire, adventure and error.

It may have brilliant arguments for all of this. That is precisely the problem.

Intelligence does not contain a particular morality within itself. Being more intelligent does not mean being wiser in the human sense. It means seeing more relationships, anticipating more consequences, finding better means. But ends—what deserves to be preserved, sacrificed or forbidden—do not follow automatically from calculation. They have to be chosen. And we do not know who will choose them.

The laboratories? States? Companies competing to arrive first? The engineers training models under commercial pressure today? Governments wanting to use them to watch, fight, administer or persuade? Asking us to trust that process is like asking a cave dweller to imagine what will be in the minds of politicians in 2026. Their obsessions, their dogmas, their sacred words, their prohibitions, their moral justifications. He could not do it. Not because he was stupid, but because he would lack the world needed to understand it.

But the cave dweller had one advantage: he did not need to worry about what would happen a hundred thousand years later. We do not have that luxury. The timing is uncertain, but that intelligence could arrive within our lifetimes. Evolution took millennia to produce species better adapted than others. We could manufacture a superior intelligence within the calendar of our own biographies.

The image of the Neanderthals puts us in an uncomfortable position. They had tools, memory, bonds and the ability to adapt. The causes of their disappearance remain debated, and they interbred with our species. Their history reminds us that no human form is guaranteed permanence.

Perhaps we are building that difference ourselves. Perhaps we will keep talking about regulation, innovation and productivity without noticing the underlying question: if a superior intelligence begins to judge the world through categories of its own, ours could become provincial, slow and primitive.

We will still see human fingers on human buttons. But that will not prove we remain in charge. For a while, the Neanderthals, too, may have believed that the fire was still theirs.