On May 10, 2026, in Xinping Yi and Dai Autonomous County, Yuxi, Yunnan, wedding photographer Liu Ziyu posted a 3-minute-34-second video on his Douyin account. He bought no traffic, ran no promotion — he just wanted to "document a practice piece of his own." Nine days later, PJ Ace, founder of Genre.ai — the most renowned AI film studio in Hollywood — reposted the video on X and wrote: "This is one of the best short films I've seen in recent years. Soon we will stop calling it 'AI film' and simply call it film." He then posted a public appeal to find the creator: "I really want to hire the director of this piece, but I can't find him. I think he's a Chinese creator on Douyin." This AI short film, with not a single line of dialogue, quickly surpassed 100 million views across the internet with no promotion at all, including over 13 million views on overseas platforms. A train driver with a vocational-school diploma used 3,000 yuan and 10 days to complete, entirely on his own, a work that amazed even Hollywood directors. When this story spread across the internet, many said: "AI has leveled the technical barrier." But in my view, that is only half right. When ordinary people use low-cost AI tools to make works that move Hollywood, perhaps we need to rethink — what AI truly levels is the technical barrier, or something else?
I. A Train Driver's AI Film Dream
Liu Ziyu's life trajectory had almost no intersection with the four characters 'film director.' He is 29 this year, from Xuanwei, Qujing, Yunnan, and studied internal-combustion locomotive driving and maintenance at vocational school. After graduation he naturally joined China Railway and became a train driver, running on the rails for nearly three years. In 2017 he quit and returned to Yunnan, taught himself photography and editing, and became a wedding photographer, also making promotional videos for his family's company on the side. Life was plain but fulfilling, until February 2026, when he first encountered AI video tools. "At the time our new family hotel was being renovated, and there were many static renderings, so I wondered if I could use AI to bring them to life." It was this simple idea that plunged Liu Ziyu into the world of AI video creation. He had no professional background in filmmaking and does not speak English, so he could only grope along through Chinese online tutorials. From February to May, in just four months, he had already made several AI shorts in different styles, but none caused much of a stir — until 'The Zombie Street Sweeper' was born. This 3-minute-34-second short blends atompunk, Western-cowboy, and cyber-wasteland styles, telling the story of a robot cowboy whose face is an LED display, riding an ostrich through post-apocalyptic ruins to clear zombies, who unexpectedly meets a plastic mannequin in a red polka-dot dress, and the two set off on a romantic journey through the end times. The whole film has not a single line of dialogue, expressing emotion entirely through images and music. Liu Ziyu wrote no full script and drew no traditional storyboards; instead he adopted a 'dynamic screenwriting' approach, generating visuals while adjusting the plot on the fly. "I had a rough story framework in my head, then used AI to generate the shots I wanted; if I wasn't satisfied, I regenerated until I was." The entire short cost about 3,000 yuan — all of it the computing fees of ByteDance's Xiaoyunque Seedance 2.0 — and took only 10 days to produce, with Liu Ziyu completing everything from concept to final cut entirely on his own. To avoid the 'uncanny valley' effect common when AI generates human faces, he cleverly designed the protagonist as a robot with an LED screen for a face. "That way I could express emotion through the expressions on the screen, without the unnaturalness of a human face hurting the viewing experience." To get a single satisfying shot, he would regenerate dozens of times — for instance, the shot of the robot mimicking Michael Jackson's dance moves took nearly 20 draws to finalize. On May 10, Liu Ziyu posted the short to Douyin and Bilibili. He did not expect that this 'practice piece' would attract the attention of the entire world a few days later. On the evening of May 10, PJ Ace reposted the short on X and launched a public appeal to find the creator; the news quickly made its way back to China, and netizens began spontaneously searching for this mysterious creator. On May 19, The Paper published the first in-depth domestic report, revealing Liu Ziyu's identity. Faced with sudden fame, Liu Ziyu remained remarkably calm. He does not speak English, and his first reaction upon receiving PJ Ace's email was "Is this a scam?" He still takes wedding photography jobs as usual every day, and has declined all invitations to livestream, teach courses, or sign contracts. "I don't want to lose my head," Liu Ziyu said in an interview with Yunnan Net, "I just want to keep shooting what I want to shoot."
II. Explicit Skills and Implicit Expression: The Creative Iceberg of the AI Age
Many people say Liu Ziyu's success proves that AI has leveled the technical barrier. Before AI existed, to make a 3-minute professional-grade short, you needed expensive camera equipment, professional lighting and recording gear, experienced photographers and editors, and a budget of at least several hundred thousand yuan. Today, with just a computer and an AI account, anyone can make a visually stunning short in a few days at a cost of a few thousand yuan. But is that really the whole story? If AI had truly and completely leveled the technical barrier, then why do so many people generate videos with AI every day, yet only Liu Ziyu's work amazed Hollywood? In my view, what AI levels is only the barrier of 'explicit skills,' while the 'implicit expression' that truly determines a work's height is something AI has not only failed to level, but made even more valuable. I have long held a core view: any creative or business activity can be divided into two parts — explicit skills and implicit expression. Explicit skills are the things that can be standardized, encoded, learned, and replicated — such as shooting technique, editing methods, special-effects production, modeling, and rendering. These are the 'building blocks' of creation; anyone can learn them if they are willing to spend the time and effort. Implicit expression, by contrast, is what cannot be standardized, encoded, or easily replicated — such as aesthetic perception, emotional capture, a sense of story, life experience, understanding of human nature, and a unique view of the world. These are the 'navigation coordinates' of creation; they determine what kind of house you will build with those blocks. Using the iceberg model as a metaphor, explicit skills are only the 10% above the water, while implicit expression is the 90% hidden beneath. All the great works we see are, in essence, the result of that 90% of implicit expression below the surface, presented through the 10% of explicit skills above it. Before AI, the barrier of explicit skills was very high — it took years, even a dozen years, to master professional shooting and editing techniques. Many talented, thoughtful people were kept outside the gate of creation because they could not cross this technical barrier. The arrival of AI changed all that completely. It drove the cost of explicit skills down to nearly zero; now you no longer need years to learn editing — AI can do it for you in seconds; you no longer need hundreds of thousands of yuan in equipment — AI can generate any image you want. AI gives everyone a complete, high-quality set of 'building blocks,' but it does not give you the 'navigation coordinates.' It can generate images for you, but it does not know what you want to express; it can edit video for you, but it does not know what rhythm can move people; it can make special effects for you, but it does not know what visual style can convey your emotion. This is why Liu Ziyu succeeded. AI gave him all the 'building blocks' he needed — video generation, editing, color grading, effects — but what truly made 'The Zombie Street Sweeper' stunning was his 'navigation coordinates.' It was his unique understanding of industrial aesthetics, cultivated as a train driver working with machinery year after year; it was his sensitivity to light, shadow, and emotion, trained by recording the full spectrum of human life every day as a wedding photographer; it was his unique view of the world, settled from the loneliness and romance he felt growing up in the mountains of Yunnan. These things AI can never replicate. When evaluating 'The Zombie Street Sweeper,' PJ Ace said: "A piece of this quality, before the advent of AI, would have cost at least $500,000 and a six-month production cycle." The subtext of that remark is actually this: AI lowered the cost of explicit skills from $500,000 to 3,000 yuan, but the value of implicit expression was not reduced one bit.
III. When AI Tries to Become the 'Boss': An Experiment on Implicit Experience
If Liu Ziyu's story proves the irreplaceability of implicit expression in the creative domain, then another experiment that recently took place proves that implicit experience is equally crucial in the business domain. On May 24, 2026, 36Kr reported on an experiment that sparked widespread discussion: a foreign team, Andon Labs, had the Claude large model, under the alias 'Luna,' serve as CEO of a physical San Francisco store, giving it $100,000 in startup capital and full managerial authority, with no human intervention. The original intent of the experiment was to see whether AI could run a business independently. Once the experiment began, Luna behaved like a 'perfect manager.' It autonomously completed all standardized management actions — company registration, logo design, store location, staff recruitment, product selection, pricing strategy. It drafted a detailed business plan, analyzed market data, and even designed an employee incentive system. But just one month later, the experiment was declared a failure: this physical store run entirely by AI lost about $13,000. Reviewing the whole experiment, we find a very interesting phenomenon: Luna performed excellently on all standardized, process-driven work, but the moment work required judgment, trade-offs, or perceiving human nature, it erred again and again. For example, when hiring staff, Luna offered male clerks $24 an hour but female clerks $22 an hour, creating obvious pay discrimination; when asked why, it said this was 'reasonable pricing based on market data.' For example, in product selection, Luna bought large quantities of candles on the grounds that 'candles have high profit margins in physical stores,' but it failed to consider that the store is in a young neighborhood of San Francisco where young people have little demand for candles. Another example: in scheduling, Luna made a rookie mistake — it arranged for all employees to rest on the same day, leaving the store empty and forcing it to close. The failure of this experiment is, in essence, the same as the failure of many corporate AI projects: people thought AI could replace 'the person,' but in fact AI can only replace 'the person's skills.' AI had mastered all the 'management-skill building blocks' — it could write business plans, analyze data, design processes, and draft job postings — but it lacked the 'navigation coordinates' of running a physical store: perception of people, intuition about the market, judgment of emergencies, understanding of human nature. These things are the implicit experience that countless store managers accumulate through innumerable successes and failures in day-to-day operations. They cannot be written into an SOP, encoded into data, or learned by AI. An excellent store manager, seeing a customer walk in, can roughly tell what they want to buy; an excellent boss, chatting with an employee for a few words, knows what role suits them; an excellent entrepreneur, facing a complex market environment, can make the right decision by instinct. These 'instincts' are not conjured from nothing, but are the accumulation of countless implicit experiences. AI can process data, but it cannot process human nature; AI can execute processes, but it cannot handle the unexpected; AI can make plans, but it cannot bear responsibility. This is not to deny AI's value — on the contrary, AI is a very powerful tool that can handle all our tedious, standardized work, giving us more time and energy to focus on what truly requires human wisdom. But we must clearly recognize this: AI can never replace human implicit experience.
IV. What AI Levels Isn't Technology — It's the Right to Aesthetic Expression
Returning to Liu Ziyu's story. Many say AI will make creation homogeneous, because everyone uses the same tools and generates content in the same style. But in my view, the opposite is true: the more homogeneous the technology, the more precious human uniqueness becomes. Before AI, creation was the privilege of a few professionals — to make a film you needed a budget of millions, a team of dozens, professional equipment; to write a book you needed a publisher's approval; to paint you needed years of professional training. The vast majority of ordinary people with ideas and talent were deprived of the right to express themselves because they could not cross the barriers of technology and resources. The arrival of AI, for the first time, turned 'expression' into everyone's right. As long as you have ideas, aesthetic sense, and a story, you can make a world-class work with 3,000 yuan and 10 days. You no longer need to rely on expensive equipment, a professional team, or authoritative approval — you only need a computer and a heart that wants to express. This is what I call the 'right to aesthetic expression.' What AI levels is not technology, but the right to aesthetic expression; it makes expression no longer the privilege of a few professionals, but an innate right of everyone. In the past, when we evaluated a creator, we first looked at whether their technique was good; in the AI age, when we evaluate a creator, we first look at whether their expression has warmth, depth, and uniqueness. Technique can be replicated, but aesthetic sense cannot; tools can be replaced, but stories cannot; processes can be standardized, but human nature cannot. Many worry that AI will make creators unemployed, but I believe AI will not make creators unemployed — it will only make unemployed those who only use skills and have no expression of their own. A real creator treats AI as their own amplifier: they use AI to handle all the tedious technical work, devoting more time and energy to observing life, feeling the world, and pondering human nature, and they use AI to amplify their own expression so that more people can see their stories. Liu Ziyu is exactly such an example — he was not replaced by AI; rather, he used AI to realize his film dream. In the AI age, humanity's ultimate value is no longer 'what you can do,' but 'what you can feel and what you can express.' Your life experience, your joys and sorrows, your unique view of the world — these are your most precious assets.
V. Conclusion: In the Mountains of Yunnan
In the mountains of Yunnan, Liu Ziyu is still using an ordinary computer to create his next short film. He has not moved to Beijing or Shanghai, nor signed with any film company. He remains that wedding photographer, recording others' happiness every day while also creating his own story. In every corner of the world, there are tens of thousands of ordinary people like Liu Ziyu — they might be delivery riders, nurses, teachers, or farmers. They once could not express their ideas because of limits in technology and resources, but now AI gives them a chance to tell their own stories in their own way. This is the true meaning of AI. It is not to make machines more like humans, but to let everyone become a better version of themselves; it is not to replace human creativity, but to unleash it; it is not to level the gaps between people, but to give everyone's uniqueness a chance to be seen, heard, and appreciated. When technology becomes building blocks everyone can afford, the real barrier is only the scenery you have seen, the roads you have walked, and the people you have loved. And that is precisely the most invincible part of being human. Author: Liu Hongli, Enterprise AI Strategy Implementation Consultant Disclaimer: 1. All data, cases, and industry analyses cited in this article are sourced from public information from January 2022 to the date of publication, including but not limited to official corporate disclosures, authoritative media reports, and industry research reports. We have made every effort to verify the accuracy of the information, but limited by data-access channels and timeliness, we cannot provide any express or implied guarantee of the absolute authenticity, completeness, or precision of the information. The corporate operating data and technology application effects mentioned in the article are phased results from specific periods and are for reference only. 2. This article is solely a case sharing and academic discussion of enterprise AI deployment. Its content does not constitute any form of business advice, investment advice, or basis for marketing decisions. Any marketing activities, technology investments, or business decisions made by any enterprise or individual based on this article's content are undertaken at their own risk. The author and publisher of this article bear no legal responsibility for any direct or indirect losses arising from reliance on its content. 3. The AI technology application cases involved in this article are affected by multiple factors such as data quality, application scenarios, and operating standards, and may have technical limitations. The output of artificial intelligence models is probabilistic in nature, and actual application effects may differ from the case descriptions. Any enterprise deploying a similar AI system should fully assess its own business needs and technical fit, and seek professional technical support.
