For a long time, making a film meant overcoming a chain of material obstacles: finding money, crew, actors, locations, cameras, shooting days and months of post-production. Production difficulty also worked as an accidental form of selection. Not everything imagined could become an image. Generative video is beginning to reduce part of that scarcity. An author can produce tests, scenes and whole short films with far fewer resources. Yet as the capacity to create expands, one resource does not: the time of the person expected to watch.
This is where the change becomes deeper than technology. If images and sequences can be generated in almost unlimited quantities, value no longer coincides with the mere existence of content. A film may be technically surprising, coherent and available at any moment, yet never meet a viewer willing to give it ten minutes. Production is no longer the principal bottleneck. Attention is.
The signals are already visible. Fairground AI Creator TV places generative cinema inside a continuous FAST channel, turning dispersed experiments into a schedule. AI festivals build selections, categories and prizes. Social platforms receive a stream of images beyond any human capacity to view. These are different infrastructures, but they share one promise: more audiovisual material will be available. None of them can make the audience infinite.
Large platforms measure this tension in hours, views, completion and retention. In its report for the second half of 2025, Netflix records 96 billion hours watched over six months. That is an immense quantity, but not a limitless space. It shows every title competing for a fraction of collective time already occupied by series, films, sport, social media, games, podcasts and everyday life. Attention has become an industrial currency precisely because it cannot be produced alongside content.
The risk is believing that the solution lies in generating still more variants. If a viewer does not choose a film, its artwork can change, its trailer can become shorter, the edit can adapt, the voice can be translated or the story personalized. AI makes all these operations cheaper. But greater precision in reaching an audience does not automatically mean greater freedom for that audience. When every image is optimized to attract the eye, cultural experience can shrink into a contest between increasingly efficient stimuli.
A film, however, is not valuable only because it can retain someone. It may demand patience, disorient us, refuse an immediate reward or become intelligible days later. Platform language tends to read a pause as abandonment and acceleration as success. Cinema knows another temporality, one in which attention is not captured but constructed. It is not always intense; it may be fragile, intermittent and even difficult. This form of availability makes an encounter with something we did not already request possible.
Infinite production therefore restores importance to curation. A festival programmer, critic, cinema, magazine or author who signs a selection does not eliminate abundance; they transform it into a path. They implicitly tell viewers that someone has watched before them, compared works and accepted responsibility for a choice. In the generative era, selection is not an accessory filter applied after creation. It becomes part of the cultural work that reaches the public.
Trust will be decisive. If the cost of producing a persuasive trailer, spectacular poster and perfect synopsis continues to fall, those signals will lose some of their power. It will become harder to know whether the promise conceals a film capable of sustaining attention. Other indicators will acquire value: who presents it, in what context, with what sources and production history; what the author has made before; why a festival included it; which community is discussing it.
UNESCO observes that generative AI automates the mass production of content and shifts the challenge from formal creation to validation. That is an essential passage. Validation does not mean certifying that a work is beautiful according to a universal measure. It means making its intention, provenance, responsibility and relationship to the culture from which it emerges legible. In an ocean of plausible images, knowing why something exists may matter as much as what it shows.
There is also a question of diversity. If attention is distributed almost entirely by a handful of recommendation systems, greater supply does not guarantee greater plurality. Millions of works can be available yet remain invisible while a few familiar aesthetics occupy most screens. UNESCO’s 2026 report warns that platform concentration and opaque selection mechanisms can marginalize lesser-known creators. Technical abundance without plural access therefore risks producing a new cultural scarcity.
Recommendation algorithms are necessary for navigating enormous catalogues, but they always embody an idea of relevance. They may favor what resembles earlier choices, what produces a rapid response or what reduces the chance that someone leaves. A human curator is not neutral, but can declare a point of view, defend a deviation, connect distant works and propose something that does not yet have enough data to be recommended. The difference is not between a bad algorithm and an infallible person; it is between invisible criteria and criteria that can be debated.
Cinema has already invented many devices for making attention valuable. A theatre separates a time and a space; a festival turns a screening into an appointment; criticism offers context; a cinematheque recirculates works the market has stopped promoting. These institutions are not analog leftovers destined to disappear. They may become more important when films are no longer rare, because they produce what automatic generation cannot manufacture by itself: a shared situation and a common memory.
Authors will also need to change their question. Not only “what can I generate?” but “why should anyone give time to this particular work?” The answer cannot simply be image quality, because that quality will become increasingly accessible. It may be a recognizable voice, genuine research, a relationship with a place, access to people and stories that are not interchangeable, or a form that uses AI to show something we could not otherwise have seen.
This does not mean turning every film into a promotional event. It means separating attention from the noise that precedes it. A work can deserve time precisely because it does not shout, pursue every trend or continuously adapt to predicted reactions. In a generative environment, subtraction may become evidence of intention: fewer images, but necessary ones; fewer variants, but chosen ones; less automatic perfection, but a form capable of preserving friction and presence.
The audience is not a passive resource to conquer. Every viewer constructs a film’s meaning through memory, expectations and emotional availability. Treating people solely as a quantity of minutes to maximize impoverishes both the work and the industrial relationship. A sustainable audiovisual culture should protect the right to choose, interrupt, return, get lost and encounter something that was not predicted by a profile.
The question “who will watch all these films?” therefore has a simple answer: nobody. Nobody can watch them all, and that is not a failure. The problem begins when production behaves as though every image automatically has a right to be seen and every free second should be filled. Abundance can be an extraordinary creative opening only if it accepts selection, silence and even invisibility as parts of the system.
The next decisive innovation in generative cinema may not be a model capable of creating longer video. It may be a set of practices capable of providing orientation: transparent credits, recognizable curators, credible festivals, accessible archives, plural recommendations and communities in which viewing once again produces conversation.
Images may become infinite. Attention remains finite, and that is precisely why it has value. The future of audiovisual culture will not be decided only by those who can generate more, but by those able to build a reason strong enough for us to stop, choose one image and watch it through.