Zilan Qian is a research scholar at GovAI researching AI development and regulation in China.
On October 21, the People’s Daily published a commentary titled “Turning ‘Gunpowder’ into ‘Fireworks’? Beware the Claim That China’s AI Has Gone Astray” (把”火药”做成”烟花”？警惕”中国AI跑偏论”). The same day, Nanfang Commentary (南方评论), the op-ed column of Southern Daily, asked: “Is Half of China’s Compute Being Spent on Feel-Good Dramas (爽剧)? Don’t Judge AI by Its Tokens.” Three days earlier, Beijing News (新京报) had questioned whether “‘American AI Boosts Productivity While Chinese AI Makes Tittytainment’ Is a False Proposition.” All three state-affiliated outlets were responding to the same viral claim: that 55% of China’s AI tokens were being used for short dramas and video generation.
The claim struck a nerve because AI-generated entertainment in China is both everywhere and widely despised. According to China’s broadcasting regulator, more than 90% of the roughly 430,000 micro-dramas released in the first eight months of 2026 were AI productions. Yet Chinese media and public figures routinely call them wunao (无脑, braindead), cuzhi lanzao (粗制滥造, shoddily made), and zhili dixia (智力低下, intellectually deficient).1
Why, in a society that places such a premium on productivity, has an industry devoted to manufacturing seemingly useless pleasure become so enormously popular? And what should we make of the shame attached to it? I argue that both answers lie in Chinese society’s obsession with usefulness, where technologies and people alike are judged by their productive value, so much so that resting can feel like a personal failing. AI slop offers an escape from that pressure, and the shame is the same pressure catching up with people even in their escape. Its shoddy production deserves criticism, but the pleasure people take in it should not have to justify itself by being useful.
“America uses gunpowder to make weapons; China uses it to make fireworks.”
In June 2026, in a meeting with investors, Chinese digital-content company COL Group 中文在线 claimed that AI short dramas and video generation accounted for 55% of China’s AI token consumption, compared with 15% for software development. The figures were picked up in October and circulated widely under the headline “China uses half of its compute to create AI slop.” Alongside this widely circulated claim is an alleged comparison showing the United States using 58% of its tokens for coding, research and enterprise automation.2 The comparison prompted some Chinese commentators to revive a familiar analogy: the West uses gunpowder to make weapons, while China uses it to make fireworks. The underlying message was that the US had harnessed technology for gains in productivity and national power, while China had squandered its technological potential on entertainment.
The analogy comes from the Chinese writer Lu Xun, who wrote in his essay “The Benefits and Harms of Electricity” (电的利弊)” in 1933: “Foreigners use gunpowder to make bullets to defend against enemies, while China uses it to make firecrackers to worship the gods.” Writing at a time when China was struggling with foreign aggression and the legacy of its technological decline, Lu Xun invoked a familiar anxiety about China’s failure to turn its own inventions into sources of national strength. Nearly a century later, the same metaphor has resurfaced in discussions of AI, suggesting that the fear of falling behind technologically remains deeply resonant.
The controversy became so intense that ByteDance, maker of the top video-generation model Seedance 2.5, publicly challenged the figures, saying that video generation accounted for only around 10% of its own computing capacity.3 The People’s Daily commentary also rejected the simplistic comparison and enumerated China’s many applications of AI in scientific research and industrial production.
How accurate is the 55% claim?
According to Chinese business media Caixin, the claim comes from a roadshow and conference call held on June 3 by COL Group (中文在线), an A-share listed digital-content company. The company’s investor relations record states that “among all token consumption in China, AI short dramas and video generation rank first, accounting for as much as 55%; followed by e-commerce and marketing (such as livestream sales and advertising) at 24%; while software development, once highly anticipated due to the popularity of vibe coding, accounts for only 15%, despite the continued advances in coding capabilities of models like Claude.”4
Caixin also claimed that the number is consistent with the Token Economy Development Research Report (2026), published on September 30 by the China Academy of Information and Communications Technology (CAICT), China’s state-backed think tank, which states that “AI short dramas account for 55% of token consumption across the entire industry.” However, the report can no longer be found on CAICT’s website.
Video generation is extremely token-heavy. On ByteDance’s own Seedance model, a five-second 720p clip costs around 100,000 tokens, roughly the length of 200 chatbot answers. Huang Ying, vice president of the token business group at Runjian (润建股份), which operates “token factories,” told World Journal on October 19 that, from her perspective, more than 50% of tokens are indeed used for video generation. On the one hand, people using generative AI for content are more active; on the other, image and video generation burn far more compute than text. Although some tokens go to R&D, R&D is essentially text, so it falls far behind video and images.
A rough calculation suggests released AI micro-dramas alone are unlikely to explain the 55% figure. According to the broadcasting regulator, 430,000 micro-dramas were released between January and August 2026, more than 90% involving AI production. Assuming, generously, that all 430,000 were entirely AI-generated and each contained 200 minutes of finished footage (20 episodes with 10 mins each), producing them at 1080p and 24 frames per second would require roughly 251 trillion video tokens. Even allowing for ten times as much generated footage as the final output to account for discarded takes and repeated attempts, the total is only 2,500 trillion tokens. China’s reported March rate of 140 trillion tokens daily would imply about 34,000 trillion over eight months, so AI micro-short dramas would account for 7.4%.
There are some uncertainties in actual production practices. Chinese creators describe repeatedly generating the same shot, sometimes dozens of times, before obtaining satisfactory results (a practice called “card-drawing” 抽卡). Yet reaching 55% from released micro-dramas alone would require roughly 75 times as much footage generated as retained—and considerably more if typical productions are shorter than assumed. The 55% figure likely combines video generation specifically for short micro-drama with other usages.
Although the original statistics lacked reliable evidence, the debate revealed an underlying anxiety about the proper use of AI. In this discourse, fireworks are inferior to weapons, entertainment to productivity, and using AI to generate short dramas to using AI to write code. Yet the disdain for supposedly useless entertainment coexists with an enormous appetite for it. As of May 2026, China had 851 million micro-drama viewers, equivalent to roughly three-quarters of the country’s internet population – and now 90% of these micro-dramas are produced by AI.
While people dismiss AI micro-dramas as meaningless and empty, they are nevertheless hooked on them. The paradox makes sense only once we see how a social obsession with usefulness pushes people toward extreme uselessness.
The Obligation to Be Useful
“Science and technology are the primary productive force” (科学技术是第一生产力), Deng Xiaoping declared in 1988. The phrase has since become a fixture of Chinese public life. Many have explained how China’s rapid industrial progress, as well as the memory of the century of humiliation, produced a deep belief in technology’s potential to make life better and the nation stronger. But this does not mean that Chinese people have an uncomplicated relationship with technology. One widely acknowledged risk is addiction. Anyone who grew up in China in recent decades will be familiar with moral panics surrounding new technologies and media. In the 1990s, it was novels; in the 2000s, it was personal computers; and in the 2010s, it was the damn phones. Online games were even nicknamed “electronic heroin” (电子海洛因). Yet the verdict always depended on how they were used. Novels were useful when they were serious classics that helped with literature exams, and harmful when they were romance or sci-fi that wasted time that should be spent studying. Computers were useful when students learned typing and coding in class, but harmful outside the classroom. These technologies were all “dual-use” in the sense that one needed to master how to use them in a useful way, while anything close to entertainment was a waste of resources. Like gunpowder, they could be made into weapons or fireworks.
AI is another example. As Beijing starts to regulate parasocial AI relationships, the signal it sent to society is that AI should be a tool for economic gains, rather than a toy or emotional companion. Even when it supports AI-generated content, the emphasis is often on how AI improves the production pipeline, generates growth, and helps boost the economy. AI slop, or sometimes called “AI swill” (AI泔水),5 is squarely outside such an emphasis and needs to be removed from the information ecosystem, as regulatory campaigns show. In essence, AI should be useful, and even when it assists in producing entertainment, the output needs to be “high-quality” content that helps people improve themselves. As Chinese media Huxiu wrote, there seems to be a ranking assessment where “production is superior to consumption, scientific R&D is better than business, and serious content is more prestigious than entertainment content.”
Technology’s obligation to be useful reflects a social mandate requiring humans to be useful. For Chinese children starting at an early age, academic performance becomes not merely a measure of ability but also of moral character. Those who excel are hardworking and deserving, while those who fall behind risk being labelled lazy or deficient. The obligation persists into adulthood, when even leisure can become another occasion for self-improvement. Chinese people have identified a phenomenon they call 休息羞耻, or “rest guilt,” in which people experience anxiety and shame about taking time away from productive activities. On the job market, a gap in one’s CV, known as a “blank-window period” (空窗期), is widely seen as a red flag; some recruiters say they automatically reject candidates whose gap exceeds six months. As a well-known joke goes, “gap year” has no Chinese translation. In a 2025 study of doctoral students, education researchers at Central China Normal University argue that effort, time, and success—originally morally neutral concepts—have become moralized, to the point that resting can feel like a personal failing. In this sense, usefulness is no longer simply a subjective quality of what one does, but a measure of who one is.
What does it mean, then, for a society where rest feels guilty to be obsessed with AI slop?
The Pleasure and Practicality of Uselessness
A familiar term in Chinese internet discourse for entertainment like AI slop is 奶头乐, or tittytainment. Commonly, though controversially, attributed to former US national security adviser Zbigniew Brzezinski, the concept imagines a future in which technological progress renders much of the population economically redundant, while cheap, endlessly stimulating entertainment keeps people pacified. The term has become a popular way of criticising short videos, online games and other supposedly mindless entertainment in China.6 Recently, people also use the phrase in the debate over China using 55% compute for AI video generation: “The US uses AI to boost productivity, while China uses it to create tittytainment.”
Yet the popularity of AI slop today presents an almost opposite picture. In the original conception of tittytainment, entertainment was needed to pacify people who had become useless to the economy. In contemporary China, however, the demand for such entertainment may arise precisely because people are under relentless pressure to be useful. Rather than having too much leisure, they have too little energy left after work and competition to pursue anything demanding. A 2024 research survey found that 63.1% of Chinese micro-drama viewers watched to relieve fatigue and relax, while 30.9% said they seek emotional comfort. 40.3% reported to be attracted to micro-dramas as they are easy to understand and do not need brain energy, while 39.6% said they like how they are short and do not occupy too much time.
Instead of a means of managing an idle population, micro-drama looks more like a way to satisfy the exhaustion of a working one by being intentionally brainless. As Chinese stand-up comedian Yu Xiangyu 于祥宇 put it: “Lofty ideas and subtle emotions take effort to appreciate. After a long, exhausting day, when you get home and collapse into bed, you don’t want some massive tome. It’s taxing on the brain, and it hurts like hell when it drops on your face. Why would I read Ordinary World 平凡的世界?7 I already lived it during the day.”
There is also something revealing about the fantasies these dramas offer. Many AI micro-dramas follow the familiar formula of shuangju (爽剧, feel-good dramas), built around a series of shuangdian (爽点, feel-good moments). And before these shuangdian were brought into dramas, there already existed shuangwen (爽文), online novels that were specifically written to serve emotional gratification. Protagonists of these micro-dramas or online novels are humiliated, betrayed, or looked down upon, only to discover that they are secretly rich, powerful, or extraordinarily talented. Plots try to maximize around moments of face-slapping (打脸) and underdog comebacks (逆袭).
Some cater more towards male readers (男频). In Battle Through the Heavens (斗破苍穹), one of the best-known shuangwen and China’s first online novel that hit one hundred million clicks, the fallen genius Xiao Yan has his engagement broken off by Nalan Yanran, whose family looks down on his decline. He returns three years later to defeat Nalan, fulfilling his famous vow three years ago – “Thirty years east of the river, thirty years west – don’t look down on a young man for being poor (三十年河东，三十年河西，莫欺少年穷)”. Micro-dramas compress the same arc that novels spent years to finish into minutes. In the much-parodied “smirking war god” (歪嘴战神) genre, a meek live-in son-in-law endures his in-laws’ contempt and even his wife’s slaps, until a crowd of men in black suddenly kneel before him to welcome the young master to take over the family group.
Others cater to female audiences (nüpin, 女频). One genre is rebirth-revenge stories (重生复仇), often set in grand historical households, where a heroine betrayed in her past life is reborn with her memories intact and methodically destroys her cheating husband, scheming stepmother, and the falsely innocent “white lotus” (白莲花). Others also go with “real/fake heiress” (真假千金) stories. They are similar to the smirking war god genre, but instead of a son-in-law, an adopted girl raised in obscurity returns to her wealthy birth family and reveals one hidden identity after another (genius doctor, elite hacker, top investor) to the shock of the snobbish relatives who scorned her.
Unlike the real world, where effort does not necessarily lead to success, these fabricated worlds offer a reassuring certainty. The good will be rewarded, the villains punished, and the protagonist will eventually have the last laugh. These plots are designed specifically for viewers who are exhausted by competition and the uncertainty of real life, offering them temporary escape from reality with fabricated, known success. Even if the plot feels unreal, the acting is bad, or the writing feels off, the audience usually tolerates these flaws because they know that these are simply for plain pleasure.
There is hardly anything more suitable for AI — something that follows fixed plots and does not require premium quality. The industry, ironically, has become extraordinarily productive at manufacturing the feeling of not having to be productive at all. The micro-drama industry had already found a lucrative business in this exhaustion, and AI made the formula that the micro-drama and online novel industry discovered even more efficient, dramatically reducing the cost and time needed to produce each episode. Meaninglessness is standardized and automated. Afra Wang argues that this AI-driven “industrial-slop complex” “has nothing to do with tech diffusion, innovation, or even entertainment,” and compares AI micro-dramas to China’s ghost cities: rows of near-identical apartments that nobody lives in. Read from the supply side, the comparison points to waste, a production machine running for its own sake.
But the analogy can also be read from the side of the viewer. In a society where every hour is crowded with competition, obligations, and other people’s expectations, meaningless slop may be the few things in people’s lives where meaning is not explicitly being asked. Yes, AI slop is like a ghost city full of emptiness. Yet this exact emptiness is what makes it attractive in a world where everything is crowded with meaning.
The Luxury of Fireworks
Of course, the emphasis on usefulness has also served China well in some respects. The country has proved especially good at adapting and scaling technologies. Despite not inventing the technology in the first place, China produces 86.4% of the world’s solar modules, makes more than 80% of the world’s battery cells, and has high-speed rail networks that now exceed 50,000 km, more than the rest of the world combined. Even in AI, where US companies have generally led at the frontier, Chinese labs have remained remarkably close and are known for their price efficiency. But the obligation to be useful can also leave little room for pursuits without an immediate, predictable return. Some of the most consequential innovations, after all, have emerged from activities whose future usefulness nobody could have anticipated.
For instance, as Chinese media Huxiu have acknowledged, the GPU, China’s most painful “chokepoint” in AI, grew out of video games: “electronic heroin,” the slop of an earlier era. “Nvidia’s early rise was inseparable from the boom in video games,” it noted. “To make 3D characters more lifelike and graphics smoother, chip engineers kept increasing the parallel computing power of graphics processors.” That same parallel computing turned out to be exactly what deep learning needed. AlexNet, the 2012 breakthrough that set off the modern AI boom, was trained on two Nvidia gaming cards.
But more fundamentally, using GPUs to justify AI slop runs into the same logic with the obligation to be useful. Saying that video games should be encouraged because it will lead to GPU is still ranking activities based on productivity. What if GPUs never became useful for AI? Would the pleasure of playing video games have been any less legitimate?
Today’s AI slop, however awful and meaningless it is, is not necessarily a sign of declining taste, nor of a nation squandering its technology. Its repetitive plots, stiff characters, and industrial production deserve criticism, and an industry that profits from people’s exhaustion should justifiably receive more scrutiny. But the pleasure people derive from watching it does not need to be measured against some greater contribution to society. Perhaps it is exactly because people are so suffocated by the obligation to be useful that the uselessness of AI slop is especially attractive.
Yet the obligation is hard to escape, and even the pleasure of doing something useless can come with guilt. Nobel Laureate Mo Yan 莫言 admitted during a panel last October that he sometimes spends hours scrolling through short videos, only to criticise himself afterwards for wasting time on largely meaningless content. His confession was greeted with relief by people who share the guilt of enjoying the pleasure derived from uselessness. “So everyone does this,” one commenter wrote. “I won’t blame myself anymore.” Yet Mo Yan still insisted that reading books is what people should really be doing with their leisure time.
Even defenders of AI micro-dramas justify the industry by pointing to its commercial value and technological spillovers: AI micro-dramas demonstrate how widely the technology is being deployed, while advances in video generation can benefit applications beyond entertainment. But does pleasure need to be economically productive to be enjoyable? As the aforementioned Huxiu commentary, “Scientific discoveries, economic growth, and higher productivity matter because, at the end of the day, we want people to lead freer, richer, and happier lives. We cannot expect AI to serve people’s livelihoods while looking down on people’s real needs.” Perhaps the deeper problem is not that people spend their evenings watching something brainless, but that even this small pleasure has to be justified.
Lu Xun’s essay on electricity is in fact about the “dual-use” nature of technology. In Shanghai, he observed, electric shocks were applied to prisoners again and again, until their teeth loosened and their nerves were dulled for good. As the city commemorated Thomas Edison, many praised the telegraph and the telephone without considering that the same electricity was being used to torture people: “The fortunate use electricity to cure illness and beautify themselves, while the oppressed suffer and die by it.” Only in the final paragraph does Lu Xun turn to how China put technologies to superstitious or self-destructive uses: gunpowder for firecrackers to worship the gods, the compass for feng shui, and opium consumed as food. Yet today’s AI debate has picked up only this last paragraph, reducing the perspective of the duality of technology into the anxiety over whether it is put to productive use. Uselessness, it seems, has become a greater worry than harm.
When Lu Xun wrote in 1933 that foreigners used gunpowder to make bullets while China used it to make firecrackers, China was in extreme crisis. Japan had seized Manchuria and attacked Shanghai, the Nationalists and the Communists were fighting each other, much of the population lived in deep rural poverty, and life expectancy was around 35 years. Today, China is far stabler and far richer, yet the sense of crisis persists. The same metaphor is now applied to the country’s AI tokens, and the same sense of obligation stigmatizes the hours ordinary people spend watching slop at the end of a long day. Perhaps one sign of a country no longer in crisis is being able to enjoy fireworks without apology. Fireworks, after all, do not have to become weapons to be worth making.
Zilan is grateful to Afra Wang and Alan Chan for their helpful feedback.
For example, TV producer Yu Zheng (于正) wrote on Weibo in June 2026 that AI dramas on the market were so poor they did not even look like real people, dismissing them as “shoddily made cartoons” (粗制滥造的动画片) without meaning. In September, veteran actor Li Liqun (李立群) publicly criticised AI short dramas for using his likeness without permission in shoddy, formulaic content that misled viewers
Most Chinese articles do not disclose sources. The number is likely taken from Menlo Venture’s “2026: The State of Consumer AI” report, which actually states that AI takes over 58% of the coding projects.
Although ByteDance claimed that video generation only accounts for 10%, that number is about computing capacity, whereas the 55% refers to the share of tokens output by models. The former measures the share of ByteDance's computing resources devoted to video generation, while the latter supposedly measures the share of tokens consumed by AI applications across China.
The mention of Claude is interesting. Many Chinese developers do use Claude and other US models, often through so-called transfer stations, which give users an API address and key and then forward their requests to the real model upstream. Since Claude’s weights are not publicly released, genuine Claude tokens can only be generated on Anthropic’s own servers or authorised cloud providers outside mainland China. These tokens are therefore unlikely to show up in China’s own token consumption records – which may partly explain why software development looks so small. (Some transfer stations do secretly swap in cheaper domestic models while labelling them as Claude, so not all of these tokens are generated abroad.)
People’s Daily claimed that swill (泔水) is the Chinese translation of slop. Although the origin of this translation is unclear, the term “AI 泔水”, and sometimes “数字泔水 (digital slop)”, is frequently used by the state.
Author Lu Yao 路遥’s novel, set between 1975 and 1985, reflects the drastic political and economic change happening in China from the end of the Cultural Revolution to the early period of Reform and Opening policies.
Zilan's posts are always such a delight to read
Also, how much of the coding in the US using AI is vibe coding for trivial hobbyist games and other nonsense?