Original · Unique Research · 2025-11-27
Historical edition note: This complete English edition preserves Unique Research’s reporting and the participants’ views as published on November 27, 2025. References to “today,” “this year,” “next year,” model capabilities and future market outcomes belong to that historical context, not a current update. Company performance, superlatives and forecasts remain attributed to the original speakers.
Editorial clarification: The claims about fast-food consumption, the percentage of repetitive work, AI replacing filmmaking, music companies’ dealings with Suno, and copyright ownership are preserved as the speakers’ historical claims, rhetorical comparisons or predictions—not independently established statistics, legal outcomes or a general statement of law. “Shuimu Intelligence” is a transliteration-based rendering of 水母智能. The source’s copyright question uses the name 李强, while its introduction and reply identify Zhou Liqiang; the English consistently identifies the guest as Zhou Liqiang.
Text visible on the original stage backdrop: “Short Dramas and Film: New Paradigms and Challenges in AI Storytelling.” The event is the 2025 Annual AI Creators Summit and CHINA AI 100 & AI CREATORS 100 Annual Selection, under the UNIQUE BLOOM theme “Pioneering intelligence — the era of the individual,” for “Founder, Creator, Builder & Influencer.” The displayed speaker names and affiliations are retained in the introduction and discussion below.
At a roundtable during the 2025 Unique Bloom in Beijing, we brought together two of today's most fiercely contested fields: short dramas and large models. The four guests onstage occupied different positions along this industrial chain.
Zhou Zhipeng, co-founder of Shuimu Intelligence, which builds AI comic dramas and content-production systems;
Liang Wei, co-founder of MovieFlow, who moved from traditional filmmaking into the AI video arena;
Zhou Liqiang, general manager of the AI Animation Department at COL Digital Publishing, which holds a vast library of IP and has invested deeply in AI animation and short dramas;
and Tong Chao, co-founder of MOOTION, which puts tools directly into the hands of individual creators;
the roundtable was moderated by Li Yong, chairman of Guyuan Cultural and Creative Technology, who has long studied emotion and content structuring.
If making a drama used to be a major undertaking open to only a small number of people, today's question has become: when everyone can use AI to make short dramas, what remains genuinely scarce? Is it the script, the language of the camera, the formula for traffic, or that uniquely human quality that is hard to define but capable of moving people?
This conversation did not stop at superficial conclusions such as “AI is powerful” or “short dramas are profitable.” Instead, it followed a more unsettling thread: as technology rapidly devours costs, labor, and barriers to entry, what can creators still hold onto so that the era does not turn both them and their work into prepackaged products?
I. When Creativity Becomes Prepackaged Food
Let us begin with a somewhat brutal reality.
Moderator Li Yong offered an exceptionally sharp view at the event: popular music is dying. Why? Because the underlying logic of popular music—the possible arrangements and combinations of chords—is finite. Once AI has exhausted every chord progression, it can continuously produce Douyin hits that sound vaguely familiar yet strangely appealing.
At that point, popular music, especially the creation of pop songs, becomes prepackaged food.
This is not happening only in music. The same logic is being replicated at a frantic pace in short dramas, novels, and even video. Once a pattern is found, AI can supply content without limit. That sounds efficient, does it not? But then comes the question: if all our food for thought becomes prepackaged fare rolling off an assembly line, will human thinking also become homogenized?
Are we training the taste buds of our minds into machines that can tolerate only a single algorithmic flavor?
This is not alarmism. Liang Wei of MovieFlow, a 16-year film-industry veteran who invested in Let the Bullets Fly, offered an even more radical judgment: in 98% of scenarios, people do not actually mind eating prepackaged food.
Just look at McDonald's and KFC: they solve the food problem for 8 billion people—a rhetorical comparison in the source, not a verified customer count. For most people, content is merely entertainment—something to snack on while doing something else. At this level, multimodal AI large models are improving at an astonishing rate. Liang even asserted that, apart from giant-screen films such as IMAX productions, AI-generated video will completely replace live-action shooting on smaller screens such as phones and televisions.
If you are an ordinary creator, this is undoubtedly a heavy blow: in an era awash in prepackaged content, mediocre creation will have no value. AI can not only do it faster than you; it may even do it better.
II. Is Peking Opera the End State of Film?
If AI takes over mediocre creation, what happens to high-end art?
Liang Wei offered the analogy that struck me most in the entire forum: the films of the future may become what Peking opera is today.
It is both a romantic and melancholy prediction. Think about it: young people in the future may watch our generation watching films much as we once watched our grandparents listen to Peking opera—a classical, ceremonial art form sustained by only a small minority.
When anyone can use AI to generate a 3D animated adventure in a Disney-like world featuring a girl, a boy, a parrot, and a monkey, the traditional film and television production process will inevitably decline. Hollywood is falling; Korean cinema is falling. It is not because they are not trying hard enough, but because technology has restructured the relations of production.
But take note: Peking opera did not disappear. It became a national treasure and a work of art.
This reveals a truth about the AI era: when technology becomes widely available, technique becomes cheap, but aesthetic judgment becomes expensive.
III. Human Compression and Decompression
So, in this era of technology racing forward, what is left for creators made of ordinary flesh and blood?
Tong Chao of MOOTION offered a compelling perspective. He believes creation is fundamentally a process of compression: you compress the flashes of inspiration and ineffable emotions in your mind into a video, a picture, or an article. An audience's reading and viewing, meanwhile, is a process of decompression—an attempt to reconstruct your state of mind at the time.
AI is exceptionally good at handling the intermediate steps, such as screenplay formatting, storyboard drawing, and visual rendering. But it cannot replace the initial instruction that performs the compression, nor can it complete the final resonance of decompression.
That is human territory.
Zhou Zhipeng of Shuimu Intelligence broke this point down even further: content consists of form and core substance. AI can exhaust all possible forms, but it cannot produce the core. Why? Because AI has no independent personality; it has never lived.
Forrest Gump moves people not because its screenplay structure is perfect, but because its writer held a distinctive and compassionate understanding of the world. AI has no childhood trauma, no pain from heartbreak, and no experience of crying late at night, so it can never generate compassion for others out of nothing.
IV. The Final Dignity: Learn to Ask Questions and Love Life
By the end of the conversation, all the guests converged on the same idea: the right to choose.
When AI can do all the dirty and exhausting work for you, the final dignity left to humanity is choice. What do you want? What do you reject? Where is your taste? Where is your judgment?
It is like learning to draw in the past: you had to spend ten years practicing pencil strokes for sketches. That was a skill of the hand. Now that AI has taken over the hand, it forces us to evolve our eyes and minds.
So, returning to the practical world, there are two extremely useful recommendations for all of us today, even if we do not work in film or television:
First, become someone who knows how to ask questions.
In the AI era, answers are cheap; questions are expensive. If you cannot think or formulate precise requirements, you can only wait for algorithms to feed you prepackaged content. Learn to use better AI models to explore the boundaries of capability instead of remaining trapped in the rules of the old world.
Second, love life passionately.
That may sound like a motivational cliché, but in the context of AI it is the most hard-core survival strategy. Liang Wei said: only when you love life can you know what AI cannot replace.
Read more books, see more of the world, and experience more real moments that AI cannot simulate. Only when you understand the world deeply enough can you give AI the most moving Prompt, or instruction.
V. A Final Note
This conversation made me less anxious about the pace of technological iteration.
The future may indeed be filled with cheap, prepackaged ideas, just as Li Yong fears. Yet, as Zhou Liqiang of COL Digital Publishing said, this is also the best time for individuals to rise. When the difficulty of production falls from 90 points to 10, anyone with even a small spark of a distinctive soul has a chance to leverage AI into enormous influence.
So do not worry about being replaced by AI. Go live; experience pain and joy. As long as you still feel the urge to express something, AI will always be the brush in your hand, not the hand holding the brush.
More Details from the Conversation
The Substitutability and Limitations of AI in Short-Drama Screenwriting
Li Yong: I will go straight to the question, beginning with Mr. Zhou. Judging from Shuimu Intelligence's work using AI to assist content creation, to what extent can AI currently replace or assist human creators in short-drama screenwriting—for example, in plot generation and character design? What are its limitations? Please elaborate a little.
Zhou Zhipeng: Hello, moderator and everyone. To summarize briefly, I think this round of AI technology creates two main forms of value in short dramas, especially in the comic-drama category on which we focus:
First, it enables high concurrency and greater efficiency in certain key roles;
second, it lowers costs and barriers in areas that require large amounts of productive capacity.
Let me use scripts and the production stages that follow as an example. At the script stage, we now use a matrix of Agents—what we call the MAS system—to help screenwriters make better adaptations. Upstream, for example, after we obtain IP from COL Digital Publishing, we first select topics with a dedicated topic-selection Agent. The screenwriter first receives a conclusion identifying the emotionally dense points in the novel's first 100 chapters that are suitable for adaptation.
At this stage, AI develops a basic adaptation strategy. Based on that strategy, it then prepares character biographies, prop descriptions, and plot distillations. It might work out how to plant a hook in the first 30 seconds, how to make users willing to pay to continue, and whether retention can be raised within three episodes. We believe AI can already provide strong support for this kind of plot construction.
Of course, there are also things it cannot do. When we previously made six-minute motion comics, AI screenwriters could largely replace human screenwriters in organizing more than 80% of the plot content. With an Agent, a human screenwriter can produce scripts for about 12 episodes a day and then make the remaining 20% of fine adjustments. But in two-minute short dramas, emotional beats are extremely dense and the requirements for hooks and conversion are higher. AI's effective contribution may then fall to 20% to 30%, requiring far more intensive human involvement.
Further downstream, in areas such as character design, image generation, and animation generation, AI delivers enormous efficiency gains in roles that once required extensive skill and experience—lead artists, screenwriters, and directors who needed 5 to 10 years of training—because it can operate at high concurrency. Image generation, meanwhile, lets people with very low barriers enter quickly: high-school students, university students, and even women in their fifties can participate, at very low cost.
The Divide Among Human Creators and the Future of Film and Television
Li Yong: I do not know how many people here regularly watch short dramas or frequently encounter work combining AI and creativity. Let me conduct a quick survey: please raise your hand if you have watched short dramas. All right, thank you. That is roughly the proportion.
Let me offer some additional context. Although we are beginning with short dramas today, our fundamental topic is how AI can help creators express themselves more effectively. Because short dramas are brief and relatively simple, AI is currently being applied to them faster and more effectively.
Mr. Zhou just raised an important point: AI can lower barriers in some stages, but other stages still require human involvement. This is a chain extending from upstream work—creativity and going from 0 to 1—through the middle and downstream stages. The real difficulty remains upstream: what is humanity's final territory? At present, AI still cannot fully replace flashes of inspiration, insight, strategy, and similar capabilities.
There is a dividing line here. Take script generation as an example. We all learned to write essays as children, and everyone in the school could do it, but the difference between good and bad writing was enormous. Today, software can produce something that looks like a screenplay in format, but is that output the same thing as a good screenplay? This is where human intelligence and machine intelligence begin to diverge.
Next, I would like to ask the two executives focused on AI applications for film and television: some people worry that combining AI with traditional narrative logic will make cinematic storytelling overly formulaic. Is that a risk?
Liang Wei: Hello, everyone. In fact, AI will not do very much in the existing market, because what we are building is an incremental market.
I may be the only person here who has spent 16 years making films. I have invested in, produced, marketed, and distributed more than two hundred films, including Forever Young and Let the Bullets Fly. As the only head of an AI-video company to have come from a film company, I can tell you very clearly: the rate at which multimodal AI large models are advancing means they can now do everything we do.
To be honest, there is little point discussing short dramas today. In one year, short dramas will no longer exist.
From my perspective, that is because everyone will be able to make them. We now have hundreds of thousands of users worldwide. A user can enter a single request: I want to create a Disney-like 3D animated story about a girl, a boy, a parrot, and a monkey; make it adventurous and Fancy, with a continuous six-minute story. In ten minutes, we generate it and deliver it in one pass.
Both China's film and television industry and the global industry—including Korean cinema and Hollywood—are declining and working for Netflix, much as music companies have had to adapt. It is like Warner and Universal in music: at first they wanted to sue Suno, the AI music generator. When they could not win, what did they do? They joined Suno.
Why did the Writers Guild of America protest when ChatGPT first emerged two or three years ago? Yet after that protest ended, do you now see the Hollywood actors' and directors' guilds staging massive protests? No. People at the forefront of Silicon Valley have seen that multimodality is advancing at such speed that its arrival is inevitable.
If I may sound the alarm, I can tell you this: by the middle of next year, apart from the giant screens used for IMAX or Cinity—which demand a level of clarity that AI cannot yet achieve—multimodally generated AI video will be able to replace live-action work on phones, other mobile devices, and televisions completely.
Li Yong: Let me press you on that point. Why does this replacement depend on screen size?
Liang Wei: Because an IMAX screen demands 4K clarity and fine detail. AI-generated video is doing well if it can currently reach 2K, and 4K is difficult. Moreover, people invest hundreds of millions of yuan in IMAX films, so that kind of production will not disappear so quickly.
Li Yong: I want to confirm your position: you say AI will replace human creation, with the distinction determined by screen size? Work at lower resolutions can be replaced?
Liang Wei: By screen size, I simply mean that in terms of clarity and the viewing experience, you will not be able to tell whether something was filmed or generated by AI.
Li Yong: Generation is a technical act, but creativity is an artistic act.
Liang Wei: AI can handle creativity too. Why could it not? In the first half of this year, many directors asked me, “Liang, AI will never be able to do a director's work.” I agreed at first. But the more we use large models, the more you realize that they are far more intelligent than you.
Humanity's final dignity is choice. The ultimate right to choose remains yours. When you use MovieFlow to make a film, for example, you still tell it what you want at the end of the process. But further ahead, just as Douyin personalizes content for each person, once AI has enough context about you, it will know what you like and create directly for you.
Li Yong: Let me see if I understand. AI is indeed more intelligent than humans and is becoming intelligent faster, so it will replace human creativity. Yet Forrest in Forrest Gump is not as intelligent as an ordinary person, and his story still moves people. As AI becomes ever more intelligent, will stories about people like Forrest—who are not clever but still move us—cease to exist?
Liang Wei: I think they will still exist, but they will become works of art. Even after the invention of photography, people continued to paint. Film may ultimately become like Peking opera: just as we watched our grandparents enjoy Peking opera when we were children, children born in the 2010s may one day watch us watching films in much the same way.
Li Yong: Thank you. This is an excellent discussion. Today, there are three broad schools of thought:
Humanity can never be replaced by AI.
As computing power, algorithms, and data achieve breakthroughs, AI will comprehensively surpass humanity and rule everything.
Human and machine intelligence will form a dual-intelligence society and coexist over the long term.
Creativity is the dividing line. Which side you believe determines how you allocate resources and where you direct your efforts next.
Copyright and Originality Protection in AI Storytelling
Li Yong: Next, I would like to ask Mr. Zhou Liqiang of COL Digital Publishing. As you develop AI animation and short dramas, do you have practical approaches to copyright issues arising from AI storytelling—to protecting originality and determining ownership of AI-generated content?
Zhou Liqiang: Before answering, I would like to continue the previous discussion. At present, COL Digital Publishing's comic dramas and live-action dramas generally involve four stages:
Story/script stage: creating and adapting a screenplay from IP.
Image-generation stage: turning the screenplay into a shot-by-shot script, designing characters and settings, and generating images.
Video-generation stage: turning images into moving video.
Post-production stage: editing, dubbing, and sound effects.
In the short to medium term, I believe humans will continue to dominate the first stage—the story—with clear differences between the best and worst work. Over the long term, however, as personalized algorithms learn to tailor content to each person, AI will also take over part of this stage. Humans will then focus more on aesthetic judgment and choice.
Returning to copyright, this is our approach:
First, at the source—the first stage—we still create and adapt screenplays from IP, resolving copyright at the story level.
Second, in visual expression—the second stage—especially for character design, we still hire artists to develop the characters before having AI learn from their work.
Overall, this is not purely generative content. Human involvement at the stages of the written story and visual design ensures ownership of copyright and mitigates the legal risks that may arise from purely AI-generated work.
Balancing Efficiency Gains and Content Quality
Li Yong: My next question is for Mr. Tong. Some say AI increases efficiency but may also trap the industry in a situation where production outpaces quality. How do you balance efficiency and quality and improve the artistry of short dramas?
Tong Chao: MOOTION sees things somewhat differently. From the first day we released the product, it was designed entirely for individual users.
We have observed that AI is enabling more people to become creators. Many people have ideas in their heads but, in the earlier environment, lacked the ability to visualize them. Tools such as MOOTION and MovieFlow now allow more newcomers to turn ideas into content that can be consumed. This is a new market, not an attempt to disrupt existing directors and screenwriters.
On that basis, I believe people will pay increasing attention to aesthetics and taste.
I strongly believe that creation is a process of compression: I visualize an idea and compress it into a piece of content; the audience decompresses it when watching, trying to understand your story.
AI is highly capable at technical tasks—screenwriting, storyboarding, and cinematography—but how do you judge what AI generates? How do you revise it? That depends heavily on human aesthetics and taste. AI, as a neutral participant, has difficulty possessing this kind of feeling.
So, although production efficiency has risen substantially, that does not mean more garbage will be produced. It means more people can produce more ideas. This decentralized, distributed mode will create more new opportunities for content.
Li Yong: More content does not necessarily mean garbage, but as volume rises, mediocre work may become more common. Most people can write essays, for example, but most of those essays are ordinary. What we want to explore is whether AI has better taste.
Zhou Zhipeng: Moderator, let me add a point. As an AI company, we are accustomed to deconstructing matters from first principles.
Content essentially consists of substance and form.
Substance: this depends on a person's independent understanding of the world. No matter how far AI advances, it will not generate an independent personality or say, “This is how I see the world.” Forrest Gump exists because its screenwriter understood the world in that particular way and therefore wrote that story. This is humanity's natural advantage.
Form: the possibilities are endless. AI is helping us solve the problem of form.
So the distinction is not screen size. It is that, at each stage, people have different needs for understanding the world, which lead them to use different forms of expression.
The Future of Prepackaged Content and Creativity
Li Yong:
Let me use popular music as an example. We have not heard many good new songs for a long time; many are simply viral Douyin tracks. Why? Because the core of popular music is harmonic progression. Once AI exhausts all the arrangements and combinations of chords, it can write viral songs that sound familiar.
But that path reaches a dead end. Once a pattern is found and recombined, the result becomes prepackaged food. It may be popular at first, then lose its appeal after a while.
If creative industries also become prepackaged food, what should practitioners hold onto?
MovieFlow co-founder Liang Wei:
The key lies in education. In the AI era, people who know how to ask questions have the greatest advantage. If you cannot think or state what you want, you will simply wait for prepackaged content to be fed to you.
In fact, 98% of people eat prepackaged food every day—McDonald's and KFC—and there is nothing wrong with that.
That makes foundational education—aesthetics, Taste, and understanding the world—even more important.
Li Yong: If this unfortunate prediction comes true and our thoughts, like our tastes, are also prepackaged—reduced to feminism or comeback fantasies—what will happen to humanity?
Zhou Liqiang: In the short to medium term, we still hope humans will intervene more in creativity itself and produce a flourishing variety of work, while handing repetitive tasks to AI.
There is a book called Bullshit Jobs, whose title refers to meaningless work. In fact, 99% of work is wasted on repetitive tasks, leaving people no time for creativity. The benefit of AI is that, if you master it, you will have more time to think creatively. This will allow people who ask better questions and have better taste to emerge.
Advice for Young People
Li Yong: Finally, please offer one piece of advice to young people or industry practitioners. What will the future look like? What should people do now?
Zhou Zhipeng: I have two trend forecasts and two recommendations.
Trends:
High-quality fragmented UGC content will explode on a large scale.
Interactive plots, open-ended branches, and deep interaction between AI NPCs and real people will become major content formats.
Recommendations:
Use AI to break through the boundaries of your knowledge and quickly master unfamiliar subjects.
Learn to collaborate efficiently with AI.
Liang Wei:
Read more books, see more of the world, and love life. Only by loving life can you know what AI cannot replace.
Use AI more—and use better AI, meaning paid large models—so you understand the boundaries of its capabilities.
Do not struggle in declining industries. Find incremental markets. Let go when it is time to let go.
Zhou Liqiang:
This is the best time to become a “super individual.” In comic dramas, for example, production difficulty has fallen from 80 or 90 points to 10 or 20. With the assistance of Agents, an individual or small team has every opportunity to produce complete episodes. Your understanding of content and professional judgment then become the core competitive advantage.
Tong Chao:
Use it more: AI is iterating extremely quickly, so keep experimenting.
Be patient: many people begin with expectations that are too high, try it once, decide it is inadequate, and give up. You must recognize that it is a trend, patiently immerse yourself in using it, and wait for the moment when a new opportunity fits you.