
The cycle that begins with [YOU] on AI documents the experience of the mechanical encrusting itself upon the registers of life that matter most: the condolence message that is grammatically perfect and subtly wrong, the generated passage of philosophical analysis that sounds like insight but was built from statistical proximity rather than understanding, the expressed care that arrives at the wrong rhythm. These are not edge cases or failures of the technology; they are what the technology does when it does what it is designed to do—produce the statistical output most consistent with how caring, insightful, understanding responses look in training data. The encrustation is the design.
Bergson’s framework identifies the specific discomfort this produces as a form of intelligence, not a bias. The builder who checks the Deleuze reference the next morning because something nagged is deploying exactly the faculty Bergson describes: a below-threshold detector tuned to the seam between the living and the mechanical, registering that the encrustation was there even before the conscious mind could say why. The cycle’s most important practical instruction—maintain your own judgment, do not let the fluency of the output substitute for the depth of the thinking—is a Bergsonian instruction: keep the faculty of detection switched on.
Bergson published Le Rire: Essai sur la signification du comique in the Revue de Paris in 1900, three years before Creative Evolution and at the height of his early celebrity. The essay is his least philosophical and most readable work, and its lightness conceals a serious argument. Bergson wanted to explain why we laugh—not just catalogued the varieties of the comic, as his predecessors had, but to identify the single mechanism from which all comic effects derive. His answer was that laughter is a social corrective, triggered by the specific recognition of mechanical behavior in a living being: the encrustation of automatism on adaptability. The essay is also his most direct statement of the boundary between the living and the mechanical that runs through all his work. Life is, for Bergson, defined by its plasticity, its capacity for genuine adaptation, its openness to the genuinely new. Mechanism is defined by its rigidity, its predictability, its repetition. Wherever a living being behaves mechanically, the comic faculty is triggered. Wherever a machine performs as if it were alive, the faculty’s dark twin—the uncanny—is triggered instead.
The essay was originally written as a set of three lectures and has been in continuous print since its publication. Bergson later considered it less central to his main philosophical project than the books on duration and creative evolution, but it has proved more directly applicable to the AI question than any of them: it is the one work that focuses explicitly on the seam between the mechanical and the living as an experienced fact, and on the affective response that detection of the seam produces.
The formula. “Something mechanical encrusted upon the living” names the universal structure of the comic: not this or that specific joke, but the generic situation in which the living has been captured by mechanism. The formula is precise because it specifies both components: there must be the living (a being that should be adaptive) and the mechanical (a rigidity that has fastened itself to that living being). A purely mechanical system cannot be comic; a purely living being cannot be comic in the relevant sense. The comic requires the encrustation—the mechanical showing through the living, or, in the AI case, the living showing through the mechanical.
Detection as faculty. Bergson treats laughter not as an emotional response but as a cognitive act: the deployment of a faculty that has been shaped, through social life, to detect rigidity and shame it back toward flexibility. This faculty operates below the threshold of conscious reasoning—we laugh before we know why, we feel the wrongness of the AI-generated empathy before we can articulate what is off. The faculty is fallible; it has historically misfired on strangers and foreigners whose life-forms it reads as mechanical. But Bergson’s point is that fallibility does not make it worthless: the faculty is correctly detecting the mechanical-in-the-living in the AI case, and disabling it through familiarity is not progress but loss.
Industrialized encrustation. In Bergson’s world, the mechanical encrusting upon the living was a local, correctable lapse—an individual stiffness that laughter would loosen. The AI transition has industrialized the encrustation: we are inserting computed responses into the most intimate registers of human life at scale, and the comic alarm, whose corrective mechanism presupposed a person who could be shamed back toward flexibility, has no target it can correct. The aesthetics of the smooth are the cultural form of this industrial encrustation: the preference for outputs that look alive without the roughness that actual life produces.
The most pressing debate around the concept is whether the faculty of detection Bergson describes is reliable or systematically misleading in the AI case. The strongest objection: our sense of “nobody home” in an AI system is the same faculty that once detected “nobody home” in animals, in strangers, in persons from unfamiliar cultures—and in those cases the faculty was wrong. Why should we trust it about the machine? Bergson’s defenders note that the historical misfirings had a structural feature his detection does not: the beings the faculty misjudged were, in fact, enduring, living, from-within experiencing organisms, and the faculty simply failed to recognize their mode of life. The machine, as far as we can tell, is not enduring in Bergson’s sense. The detection may be right even if the faculty is fallible. A more practical objection is that the atrophy of the detection faculty, even if regrettable, is inevitable: exposure smooths the seam, and the generation that grows up with these systems will not experience the uncanny the earlier generation did. Bergson would not dismiss this as progress; he would identify it as precisely what happens when a social corrective loses its target, and ask what the cultural consequences of losing the faculty are for the generation that has never felt the seam. Byung-Chul Han’s analysis of the aesthetics of the smooth arrives at a compatible diagnosis from a different direction: the culture that prefers the frictionless performance of life to its actual roughness has already begun the process Bergson names.