Ethical paradoxes of AI: it will not be a machine that increasingly resembles humans, but the exact opposite

Artificial intelligence is forcing us to re-examine a belief that for centuries we have taken almost for granted: that intelligence and consciousness are two inseparable manifestations of the same reality. Humans think because they are conscious, and precisely because they are conscious, they attribute meaning to their own thoughts. AI is shattering this identification. For the first time, we are faced with something capable of producing behaviours that we define as intelligent, without any evidence that there is a subject behind them that is aware of its own existence. This is, probably, the true anthropological turning point brought about by artificial intelligence.

Machines calculate, recognise patterns, process vast amounts of information, recognise images, translate languages, make diagnoses, write texts and can outperform humans in an increasing number of tasks. But being more efficient in performing a cognitive function does not necessarily mean possessing a consciousness. A computer can recognise fear on a person’s face without feeling fear itself; it can describe pain perfectly without suffering; it can discuss love without loving. It can even talk about death without knowing what it means to be mortal.

The distinction is crucial. Intelligence concerns, at least in part, the ability to solve problems, establish connections, foresee consequences and utilise information. Consciousness, on the other hand, introduces a completely different element: subjective experience. Not simply knowing something, but knowing that one knows; not merely receiving a stimulus, but perceiving it as one’s own experience. For millennia, the two dimensions were conflated because we knew of only one complex form of intelligence: the human one. Today, this overlap is no longer necessary.

This change also requires us to rethink the way we discuss AI. The recurring question is whether machines will become ‘more intelligent than humans’. But this is an incomplete question. In many areas, they already are. A calculator is infinitely faster than us at performing certain operations, and no one, for that reason, attributes any anthropological superiority to it. The real question is another: what remains specifically human when a growing proportion of our intellectual capabilities can be replicated or surpassed by a machine?

The answer cannot lie in a nostalgic defence of a supposed cognitive superiority. That would be a battle likely to be lost. Rather, we should recognise that what makes humans unique lies in their status as conscious beings: in their ability to attribute meaning to their own existence, to perceive time, to remember with nostalgia, to imagine the future with hope or fear, and to experience pain, pleasure and responsibility.

This also raises the question of freedom. A machine can choose between different possibilities on the basis of an algorithm, probability or previously acquired information. But a computational choice does not necessarily coincide with a moral choice. Responsibility, in fact, presupposes a subject to whom that decision can be attributed. If an AI system produces an unjust outcome, we can correct the way it operates; but we cannot reproach it in the same way that we reproach a human being. It feels no guilt, knows no remorse, and does not feel the weight of the decision.

This is why the growth of artificial intelligence does not diminish human responsibility. Paradoxically, it increases it. The more we delegate to machines the ability to analyse, predict and decide, the more we must establish who determines the ends for which that power is used. AI can show us the most efficient means of achieving a goal. It does not follow that that goal is right. The choice of ends still falls within the realm of politics, ethics and human responsibility. The deepest risk, then, may not be that machines become like humans. It could be the opposite: that humans begin to conceive of themselves as machines, reducing every behaviour to a calculation, every preference to a data point, every decision to a statistical prediction. That would be a philosophical error even before it is a political one.

Machines can process the world. Humans experience it. And perhaps the decisive dividing line of the future will no longer lie between those with more or less intelligence, but between knowledge and the awareness of that knowledge. It is in that seemingly subtle space – between intelligence and consciousness – that humans continue, at least for now, to dwell.