The first AI theatrical film "Sanxingdui: Future Past" has a cold box office reception; People's Daily comments on the challenges of AI creation
The first domestic AI theatrical film "Sanxingdui: Future Past" has seen a dismal box office since its release, with audiences criticizing it as "AI at a glance, lacking a sense of real people, and showing off technique more than content." The film follows the "visuals-first" production logic of the earlier 13-episode AI short drama "Sanxingdui: Future Revelation," and its story lacks meaningful connections. The article points out that when the marginal cost of AI spectacle approaches zero, the ceiling of AI films depends on the parts AI cannot do: script, performance, emotion, and judgment.
Topics
Body · AI translation
Since its release, China's first AI (artificial intelligence) theatrical film, "Sanxingdui: Future Past," has fared poorly at the box office, and viewers' criticisms have also been strikingly consistent: it looks like AI at a glance, lacks any sense of a living person, and values showing off technique over content. The reasons for its failure concern not only the quality of a single film but also reflect the more general challenges and problems of AI creation, and are therefore worth discussing.
From AI short videos and AI short dramas to AI films, the boundaries of film and television creation are being continuously expanded by new technology. Especially since the beginning of this year, more than 10,000 AI animated dramas alone have gone online in China each month, which means the marginal cost of obtaining AI spectacle has approached zero. This therefore raises a fatal question: when AI is almost infinitely available and within easy reach on mobile phones, why should viewers still buy tickets and go to the cinema to watch an AI film?
Or let us put the question another way: today, when viewers spend several dozen yuan on a movie ticket, what exactly are they expecting? There are many answers, but in essence it is a stretch of time that cannot be paused. AI spectacles freely scrolled through on mobile phones can be paused, sped up, or swiped away at any time; the cinema is exactly the opposite, since leaving carries a cost. When viewers enter a cinema, they are in fact taking on a certain risk, and the film must compensate them with a unique experience. That compensation is the temporal structure that "only works if watched continuously": every frame builds momentum for the plot that follows, the foreshadowing planted earlier is paid off at the end, and viewers can enjoy the entire story's setup, development, climax, and resolution in full.
"Sanxingdui: Future Past," by contrast, brought the logic of short dramas into theatrical release. The producers had previously made a 13-episode AI short drama, "Sanxingdui: Revelation of the Future," which launched for free on a platform two years ago and drew considerable clicks across the internet. According to the creators, this film also first generated a large number of images and then reverse-engineered part of the story from them. This is clearly not the production logic of a feature-length film. With images coming first, the underlying logic is flawed, the images lack meaningful connections among themselves, and the story is therefore not well told. Netizens pointed out sharply that it "looks like several different group assignments pieced together."
A deeper problem lies in the fact that the film took several years from project approval to release, and the speed of revision could not keep up with the speed of AI technological iteration. In particular, AI has already raised viewers' aesthetic threshold. Blockbusters that once cost huge sums to build practical sets for are now instinctively felt by viewers to be something AI "can do just as well." When technological spectacle becomes commonplace, the true upper limit of AI films points to what AI cannot do: scripts, performance, emotion, and judgment. In the end, the upper limit of humans determines the upper limit of AI.
Placing "Sanxingdui: Future Past" in the broader context of this year, its experience is not an isolated case. Some AIGC (artificially generated content) works bearing the "first" label also attracted attention when they first aired, but were not entirely satisfactory after broadcast. Of course, the failure of individual works does not mean that the path of AI film and television is unworkable. In the long run, low cost comes first, then enough attempts, and then good works are screened out; AIGC will expand the entire audiovisual industry. In the final analysis, AI should be a paintbrush that empowers imagination, not a gimmick that conceals empty content. Letting technology truly serve the essence of art and move people through emotion is the only true path for creation.