Fan Bingbing says young viewers are underfed. Blame the kitchen

The easy reading of China's AI microdrama boom is that young people have bad taste and cheap thrills are what they like. I think that reading is backwards. A flood of cheap, machine-made drama is what you get when the kitchen finds it easier to make one dish than ten.

That is my argument, and it starts with a remark from an unlikely place. At a press conference at the Busan International Film Festival, the Chinese actress Fan Bingbing, who was there to present Wayne Wang's "Diary of a Mad Old Man" and Ivan Tan's short "She Said," was asked about screen content in China. She said young people there "aren't being fed very well." ^[1]

She did not say it with contempt. Fan told the room she watches AI-generated microdramas herself, occasionally. What she doubted was whether they give young viewers the artistic nourishment she got from cinema. ^[1]

What she actually pointed at

Fan's comparison was about performance. Human acting, she said, carries many emotions and expressions, and the work in Lee Chang-dong's "Possible Love" is something AI could neither convey nor replicate. ^[1]

Then she offered a detail that does more work than any statistic. Park Chan-wook's "The Handmaiden" has never been released in China, she said, and she had to make a special effort to find a way to watch it. ^[1]

Notice the verb. She did not say she chose it over something else on the menu. She said she had to make an effort. If a film star has to go out of her way to see a film she plainly admires, I doubt a young viewer scrolling a phone feed finds it any easier, or even knows what is missing.

I should be careful about how far that anecdote travels. One unreleased film does not explain a vast industry, and nothing in the record shows that restrictions on foreign releases caused the AI boom. Nor do we know what young viewers would choose if the shelves were fuller. What the evidence does show is the shape of the supply, and the shape is odd.

The numbers on the plate

At a September briefing, China's National Radio and Television Administration, the NRTA, said 430,000 microdramas went online in the first eight months of 2026. That is 13 times the total for all of 2025. More than 90 percent of this year's titles were AI-generated. ^[3]

The audience is real. The same official said microdrama users had passed 800 million and that the market was worth more than 100 billion yuan in 2025. ^[3] Another industry body, the China Netcasting Services Association, estimated that more than 95 percent of the roughly 128,000 short dramas released in the first quarter were AI-generated. ^[2]

The obvious reading is demand: people want this, so people get this. But demand does not make hundreds of thousands of anything. Producers do. And producers respond to cost.

Chinese teams can make an AI microdrama for an estimated $10,000 to $20,000, according to Lin Qiwen, a vice-president at the analytics firm DataEye. A live-action season costs about $300,000. ^[5] That is a gap of 15 to 30 times. One AI series, "Zhanxiantai (Real Person AI Version)," drew more than 100 million views in six days, and its production team said it cost 100,000 yuan in computing power and a month of work. ^[4]

Now ask what a sensible producer does. CNBC reports that short dramas work as relatively low-capital bets: make a batch, see what the audience does, and put promotion money behind the winners. ^[2] When testing costs this little, you test constantly, and you test whatever is fastest to make.

The best case for the flood

The format deserves its due. Microdramas run a few minutes an episode and are commonly shot vertically for the phone. ^[5] Sensor Tower's Seema Shah told CNBC they offer a quick hook and ask little commitment of the viewer. ^[2] The Malaysian actor Kevin Chen says the short running time suits brief stretches of leisure. ^[5] Anyone who has watched a commute or a queue disappear into a phone knows the logic.

Industry figures also argue the format lowers barriers to entry, since it can be made with modest budgets, smaller crews and accessible distribution platforms. ^[5] That is a real good. A cheap format lets new people in, and many careers have started as someone's cheap experiment.

But the case for cheapness has limits that the industry's own reporting exposes. CNBC notes that promotion costs can erode the savings from cheap production. ^[2] And the same report quotes Shah saying audiences still value high-caliber, well-produced content. ^[2] If viewers truly wanted only the quick and the hollow, no one would need to say so.

The other side of the ledger is people. Chen Yuxi, a microdrama actress, found her casting group chats silent after the 2026 Lunar New Year holiday and received no job offer in February. She learned that many peers had moved to AI shorts or left the business. ^[4] That is one performer's account, not a measure of the whole sector. But it shows what a cost gap can do in practice. The cheaper machine does not just add a dish to the menu. It can starve the cooks who made the other ones.

Researcher Zhao Hui puts the creative complaint bluntly, according to Beijing Review: many current AI dramas have spectacle but no stories, visuals but no soul. Viewer Lin Huan describes characters who look realistic but sometimes have unfocused eyes and unnatural expressions. ^[4] Fan's point about human faces is the same point, made by someone who has spent her life in front of cameras.

A regulator that says the right words

In fairness to the authorities, they are not pretending this is fine. The NRTA official said AI dramas must be labelled, platforms must respect portrait and voice authorization, and platforms must conduct full-process vetting. He said 68,000 microdramas that violated rules had been taken down across the internet that year. ^[3]

He also said the regulator planned to support premium microdrama creation and live-action sections on platforms, and that live-action dramas remained the main force in premium creation. ^[3]

That is the right instinct, and I would not mock it. But look at what the toolkit is. Labels, authorization rules and takedowns are ways of policing what is already being made. A live-action section on a platform is shelf space, which helps only if something is stocked on it. The hard part, getting more ambitious and varied stories financed, cleared and made, is the part the briefing describes least.

I spent twenty years as a foreign correspondent, and what I took from it is that people make do with what is within reach. They are not lazy and they are not stupid. They eat what is on the table, and they are grateful for a hot meal. Then someone with a platform tells them the table is the full extent of their taste.

What this is really about

This is not a Chinese problem alone. Every large platform in the world faces the same temptation. A format that costs a fifteenth to a thirtieth as much to make, can be tested overnight and fills a spare three minutes will win the spreadsheet every time. The result is a diet built to the producer's convenience and then defended as the customer's choice.

Fan's phrase, "aren't being fed very well," is gentler than that. It blames no one and it names a fact of life. If she is right, the remedy is not scolding the people on the phone. It is a supply of work worth the trouble: human faces, longer arcs, stories that do not have to be hunted down the way she hunted down one film.

What to watch is whether the NRTA's promised support for premium and live-action work turns into money and visible shelf space, or stays at the level of a briefing. A regulator that took down 68,000 dramas in a single year clearly has the reach. It can surely find room on the menu for something people are not merely scrolling past.