Free Will, or Algorithmic Surrender?

How do you define freedom? Is it the existence of options, or is it ownership of the decision? And more pointedly: are you actually the author of your decisions today, or merely exposed to their outcomes? Are we deciding, or are we being decided upon?

If we define ourselves as a 'subject' in the classical sense, the model is straightforward: a person weighs alternatives, decides, and takes responsibility for what follows. That chain is now breaking apart, link by link. In most everyday situations, what actually reaches us are options that have already been filtered by an algorithm, and the decision itself is systemically steered toward one of them. The cognitive labor of choosing is migrating away from the person and into a system — fragmenting there, and becoming harder to trace.

The first fracture is epistemic. Deciding well requires access to the knowledge a decision rests on. But in modern systems, the link between information and decision is severed. A credit score, a hiring algorithm, a recommendation engine each produce an epistemic structure whose internal workings we do not understand and, in most cases, cannot access even if we wanted to. Jürgen Habermas built his model of public reasoning on the premise that decisions should be arguable and justifiable — open to challenge, traceable to their grounds. Algorithmic decisions, by contrast, are frequently unexplainable, undebatable, and untraceable after the fact. This renders the individual epistemically dependent: one goes through the motions of deciding, while having no real access to the basis on which the decision rests.

The second fracture concerns responsibility. Modern ethics rests on a direct link between an action and the person accountable for it. Algorithmic systems dissolve that link. When an algorithmically shaped outcome causes harm, the question of who is responsible becomes genuinely difficult to answer. Is it the engineer who wrote the code? The institution that deployed it? The user who accepted its suggestion? Whoever curated the training data? In practice, the honest answer is: no one, fully — a little of each. This is the systematic distribution of responsibility. And responsibility that has been distributed across enough actors tends, in practice, to disappear altogether. No one carries enough of it to be held to account, and so accountability itself becomes a kind of fiction.

The third fracture is about where rationality lives. Max Weber read the modern world through the lens of rationalization — the steady replacement of tradition and impulse with calculated, rule-governed action. What we are living through now is not simply more of that same process; it is something structurally different: the externalization of rationality. The human being is not becoming more rational. Rationality itself is being handed off to systems. The logic behind a decision no longer resides inside the deciding subject — it resides inside the infrastructure that produced the options in front of them. Gradually, the human stops being a rational decision-maker and becomes, instead, a being adapted to rational systems: optimized for compatibility with the machine's logic rather than practiced in exercising judgment of its own.

The fourth fracture is psychological, and it is the one that implicates us most directly, because it cannot be blamed entirely on the technology. Daniel Kahneman's research on decision-making shows that choosing under uncertainty carries real cognitive costs: stress, the burden of weighing incommensurable outcomes, and the risk of regret if the choice turns out badly. Algorithms do not actually remove this burden. But they appear to. And that appearance is enough to produce a felt, if illusory, sense of relief. Letting the system decide, or at least narrow the field so heavily that a decision feels obvious, offers something people quietly want: less weight to carry.

Taken together, these four fractures point toward a question that is not technical but ontological. If a person no longer decides in any meaningful sense, cannot access the knowledge behind the decisions attributed to them, and does not fully bear the responsibility for the outcomes — can that person still be called a 'subject' in the classical sense? And underneath that question sits an even sharper one: does the human being actually want to keep being that kind of subject? Because we want the freedom to choose. We do not, in practice, want to carry the full weight of choosing. We hand over the authority to decide voluntarily, in exchange for comfort. Which means the deepest issue here is not really the threat posed by artificial intelligence. It is, once again, a human preference — one that predates any algorithm and that algorithms have simply made extremely convenient to indulge.

This reframing matters because it changes where responsibility for a remedy should sit. If the problem were purely technological — opaque systems imposed on unwilling subjects — the fix would be mostly regulatory: better transparency requirements, explainability mandates, audit trails. Those remedies are necessary, but they are not sufficient, because they do not touch the psychological economy that makes algorithmic surrender attractive in the first place. A fully transparent, perfectly explainable system can still be one that people prefer to defer to, because the appeal was never solely about opacity — it was about not wanting to carry the decision's weight. Any serious response, then, has to work on two fronts simultaneously: building systems whose logic can be inspected and challenged, and cultivating, in individuals and institutions alike, a willingness to reclaim the discomfort of deciding — because that discomfort, uncomfortable as it is, is also where responsibility, and therefore agency, actually lives.