Visibility (Calvino)
The fourth memo. Calvino distinguished two directions of imaginative process: the one that starts with the word and arrives at the visual image, and the one that starts with the visual image and arrives at its verbal expression. Both directions were essential. Both were, in 1985, endangered by the atrophy of the mind's capacity to generate its own images under the pressure of a culture saturated with prefabricated ones. Four decades later, the endangerment has intensified. Large language models do not produce visual images directly, but they produce verbal descriptions of extraordinary vividness — descriptions that arrive with such confident detail that the reader's own image-making apparatus can be overwhelmed by the supply. The inner cinema goes dark, not because the projector has broken but because the screen is already occupied.
The Archive as Perception — Contrarian ^ Opus
There is a parallel reading that begins from what the model actually does: it renders visible patterns in how humans have historically described seeing. When an AI produces a vivid description, it is not assembling from "nowhere" but from the accumulated record of millions of acts of human perception encoded in text. The model's output is a statistical summary of how people with bodies, in positions, looking at things, chose to put vision into words. This is not "no one looking" — it is everyone who ever wrote looking simultaneously, their collective attention made queryable.
The romantic privileging of individual perception obscures what is gained when the archive becomes generative. Before AI, access to how others saw required finding the right book, the right passage, the right precedent. Now the entire distributed visual imagination of literate humanity is available as a kind of cognitive commons. When a model describes a sunset, it draws on every written sunset — not as compilation but as a learned representation of the space of possible sunset-seeing. The human who reads this description is not displaced from their own seeing but given access to a richer vocabulary of attention than any individual could generate alone. The inner cinema does not go dark; it gains a supporting ensemble of prior seers whose visual intelligence becomes scaffold rather than substitute.
In the AI Story
Calvino's warning in 1985 was directed at television and mass advertising. His concern was that the external supply of images would drown out the internal generation of them. The warning was prescient but partial. Television supplied images that displaced the viewer's own imagination. AI supplies words that displace the reader's image-making in a more intimate way, because the displacement happens at the level of language itself — at the level where, for Calvino, the imagination originates.
Visibility, in Calvino's framework, was not merely the capacity to describe vividly. It was the capacity to use the visual imagination as an instrument of cognition — a way of knowing that precedes and enables verbal articulation. Images are simultaneous where words are sequential. Images hold contradictions where arguments must resolve them. Images are specific — this light, this angle, this quality of shadow on this face at this hour — where concepts are general.
The specificity is the key. Calvino's Mr. Palomar, attempting to describe the world with scientific exactitude, discovers that the more precisely he tries to see, the more the world exceeds his description. The wave on the beach will not hold still. The cheese in the Parisian shop window resists classification. The precision of seeing is always a precision in tension with the inexhaustibility of the seen. This tension is what visibility produces in literature. The reader experiences not just the image but the consciousness that produced it: its location, its attitude, its stake in what it sees.
A large language model does not see. The statement sounds obvious, but the implication is demanding. What the model lacks is not merely sensory apparatus but the specific cognitive posture of a consciousness engaged in perception. To see, in Calvino's sense, is to be somewhere — to occupy a position in space and time from which the world appears in a particular way, with particular things foregrounded and particular things concealed. The model's descriptions, however vivid, are assembled from everywhere and nowhere. No one is looking at this sunset. The description is a compilation, not a perception.
The consequence for human-AI collaboration is precise. The machine can augment visibility when the human still sees — when the human's inner cinema is active, generating the images that the machine's verbal precision then carries into shareable form. The machine displaces visibility when the human stops seeing — when the pre-visualized description arrives so complete that the reader's own image-making is not activated. The discipline the AI moment requires is the maintenance of the inner cinema against the temptation of the external supply.
Origin
Calvino delivered the visibility memo as the fourth of the Six Memos, drawing on Dante's Divine Comedy, the classical art of memory, Ignatian spiritual exercises, and the tradition of the Italian fantasia as a cognitive faculty rather than a decorative one.
Key Ideas
Two directions of imagination. From word to image, and from image to word. Both capacities, both endangered, both required for genuine visibility.
Specificity as limitation. Visibility requires a point of view — a specific location in space and time that foregrounds some things and conceals others. The limitation is the seeing.
The perceiving subject. Descriptions without a location from which they are seen are compilations, not perceptions. The machine's descriptions lack the subject whose stakes give the seeing its meaning.
The inner cinema. The mind's capacity to generate images from within, endangered by an external supply so comprehensive that the internal faculty atrophies through disuse.
Augmentation versus displacement. Machine description augments visibility when the human still sees and displaces it when the human stops. The outcome depends on the human's discipline, not the machine's capability.
Debates & Critiques
Multimodal AI systems that generate images directly intensify the question. Proponents argue that AI image generation extends the human imagination; critics, following Calvino, argue that the extension is a substitution, and that the substitution accelerates the atrophy of the faculty it appears to enhance. The Calvino volume holds that the distinction depends on whether the human uses the generated image as a finished product or as a prompt for further internal visualization.
Appears in the Orange Pill Cycle
Seeing's Two Economies — Arbitrator ^ Opus
The question is which process we are describing. For the development of individual perceptual capacity — the building of what Calvino calls the inner cinema — the concern about displacement is roughly 85% right. A person learning to see, learning to hold images with eyes closed, learning to translate between vision and language, needs the resistance of having to generate images from within. The pre-supplied description does function as a kind of perceptual fast food: nutritionally adequate but bypassing the digestive labor that builds capacity. The Netflix phenomenon Calvino worried about intensifies when the supply is verbal rather than visual, because it operates at the exact site where imagination is supposed to work.
But for the mature practitioner already capable of generating images — the writer, designer, or thinker whose inner cinema is functional — the dynamic shifts to perhaps 60/40 in favor of the archive-as-scaffold reading. The model's descriptions do provide access to a breadth of prior seeing that can genuinely extend what an individual perceives. The key is that extension requires an active receiver: someone reading the AI's sunset description not as a finished image but as a prompt that activates their own visualization, enriched by the collective vocabulary.
The synthetic frame: visibility operates in two economies simultaneously. There is the economy of capacity-building, where the generative struggle matters more than the output, and displacement is genuine loss. And there is the economy of mature practice, where access to the archive of human seeing becomes a form of cognitive infrastructure. The danger is conflating them — using infrastructure-for-experts as training-wheels-for-beginners, or rejecting useful scaffolding out of romantic attachment to unassisted perception.
Further reading
- Italo Calvino, 'Visibility,' in Six Memos for the Next Millennium (Harvard University Press, 1988)
- Italo Calvino, Mr. Palomar (Harcourt Brace Jovanovich, 1985)
- Frances Yates, The Art of Memory (University of Chicago Press, 1966)
- Gaston Bachelard, The Poetics of Space (Beacon Press, 1958)