A poster can promise a technological first. It is not enough to fill a cinema. Released in China on 23 October 2026, “Sanxingdui: Future Past” puts AI-generated cinema to a less spectacular test than an image demonstration: persuading an audience to spend one hundred minutes on a story.
The first reports, covered by France Inter and Télérama, describe a difficult launch. They shed light on a specific industrial experiment. They do not allow us to conclude either that every work made with AI will disappoint, or that the film professions have already disappeared behind a generation button.
A feature film, with a team behind the images
The project is backed by Bona Film Group. In its presentation published before the release, AWN China describes a film of around one hundred minutes, produced using the Boka Cloud Film Studio system. The article names Yue Ming as screenwriter and director, Jiang Defu as chief producer and Quji Xiaojiang as executive producer.
This human presence is not a secondary nuance. The project’s presentation credits the team with the concept, narrative logic and artistic choices; AI tools are involved in visual generation and technical execution. The expression “film entirely generated by AI” therefore refers to a method of producing audiovisual elements, not to an absence of human decisions.
AWN China presents the characters as original digital figures, without reproducing real actors. This is a feature announced by the project, not an independent legal assessment of all the rights involved. The promotional wording should not, by itself, be treated as a general certification.
Chinese archaeology transformed into science fiction
The story links the Sanxingdui ruins to a future dominated by artificial intelligence. According to the synopsis presented by AWN China, an archaeologist discovers in ancient symbols a key to answering a crisis affecting humanity. The objects and motifs of the ancient Shu civilisation thus become material for fiction.
This distinction must remain clear: a script that gives futuristic codes to ancient remains is not an archaeological discovery. The project’s interest lies here in bringing together a heritage-inspired imaginary and digital production, not in offering a new scientific interpretation of the objects.
The film also puts AI on both sides of the screen. It is used to create the images and becomes the subject of the story. This choice makes the promise particularly exposed: the audience is invited to judge both a story about technology and a work that presents that technology as a reason to launch it.
A weak start, read in light of the dates
In its article of 26 October, Télérama reports takings of 1.22 million yuan, according to ticketing platform Maoyan, for the first few days after release. This figure describes one stage of the film’s run in cinemas. It should not be presented as the film’s final takings, still less as a complete assessment of its profitability.
On 27 October, France Inter reports a mixed reception and gathers several reactions from cinema-goers. Some distinguish relatively accomplished technical execution from weaker engagement with the story or the characters’ performances. These are individual observations gathered by the radio station: they do not measure the views of the entire Chinese public.
Their significance lies elsewhere. They show that judgement is not solely about whether a visual anomaly is present or absent. An image can be technically coherent without making a scene moving; a succession of effects can impress without making people want to follow the characters. These are questions of cinema that remain when the production method changes.
Technical novelty does not replace connecting with the audience
Télérama reports that Jiang Defu cited low promotional investment to explain the launch. This is the explanation put forward by the executive, not a cause demonstrated in isolation. A film’s visibility, its exposure in cinemas and audience reactions can combine; the information available does not make it possible to assign a precise weight to each of these factors.
The same caution applies in the other direction. Turning this launch into definitive proof of AI’s artistic failure would claim far more for the early takings than they show. A commercial result tells us about the meeting between a film, its screening schedule and its audience at a given moment. It does not settle all the future possibilities of a technique.
One concrete lesson nevertheless remains: announcing an unprecedented process does not remove the need to create something the audience wants to see. Curiosity about how it was made may lead people as far as the trailer; it guarantees neither the purchase of a ticket nor satisfaction on leaving the cinema.
Three criteria for watching what comes next
The first is artistic: how does a work organise its characters, scenes and pace? This judgement requires access to the film. This article analyses the launch and the production information; it does not claim to replace a review based on having seen the film.
The second is economic: which costs and revenues genuinely belong to the project? A box-office figure is not enough to reconstruct all the costs, revenues or how they are shared. Claims of cheaper production should therefore be examined using comparable data.
The third is professional: who writes, chooses, assembles and takes responsibility for the result? The credits for “Sanxingdui: Future Past” already provide a partial answer: teams remain at work. The useful question for what comes next will be how their work evolves, rather than erasing them behind the label “100% AI”.
Illustrative photograph: the entrance to the ZChow cinema in Beijing in August 2025. The posters on display are not for “Sanxingdui: Future Past”.
