But what happens when these signals no longer come from a body, but from code? A voice assistant anticipating our needs. A generative system responding with apparent relevance, or apologising with a tone of regret, almost guilt. An algorithm deciding which content to show us to keep us hooked. Are these merely tools, or are we witnessing — and contributing to — the birth of a new form of intentionality, more subtle, distributed, and shared?

Traditionally, intentionality is understood as the capacity of a mind to be about something — to desire, to think, to remember. It’s what distinguishes a mental state from a mere physiological event. But when we try to grasp it conceptually, it slips away: it’s direction, not content; structure, not substance. Daniel Dennett, American philosopher and logician, offered us a conceptual shortcut that fits this problem rather well: intentionality is not an intrinsic quality of a system, but a useful perspective. It is something we attribute to an entity whenever it benefits us to do so. A thermostat, in a sense, “wants” to maintain a constant temperature. An automated chess player “seeks” the best possible move.

Not because they feel anything, according to this shortcut, but because attributing a purpose to them helps us understand them, predict them, use them. The intentional stance, as Dennett calls it, is a form of functional fiction. And like all well-crafted fictions, after a while we cease to perceive it as such.

If intentionality is, at least in part, a product of the observer’s gaze, then there exists a grey zone where the boundary between what is and what appears begins to dissolve. When we observe coherent behaviour — human or non-human — we instinctively attach an intention to it. We cannot help ourselves. It is a cognitive anchor: we attribute purpose even to things that have never had it. And it’s here that simulations begin to challenge us. An artificial system can learn how to respond to linguistic stimuli, generate congruent sentences, even express contradiction, irony, hesitation. But these responses, theoretically, do not derive from any internal state. There are no desires, beliefs, or fears. Only calculations, patterns upon patterns. And yet, we read them as signs of interiority. We interpret them as intention. The machine has no intention, but it functions as if it did. Which, in a sense, ought to be enough.