Original · Unique Research · 2026-04-14
Editor's note: The first-person report and its judgments belong to the original Chinese author. This English rendition retains the opening essay and full panel, including all four named question rounds and every speaking turn. Claims about art, copyright, technology, companies and market conditions are source or speaker claims, not independently audited findings. Company, personal and work titles are transliterated where official English forms remain unverified. The "snow mountain fox" video attributed to OpenClaw, the Duchamp Fountain reference, and comparisons to The Wandering Earth, Ne Zha and Minecraft are preserved as speaker statements without independent verification.
Unique Awards
When AI Becomes the Brush, What Is Left for the Creator?
When technology becomes thoroughly tool-like, what gets redefined is not only the boundary of art, but also "who counts as an author" and what humans can ultimately take responsibility for in a work.
"
In the past, the brush was a physical object, dipped in pigment. Today, the brush has become code and models, dipped in data.
At this roundtable during Hangzhou AI WEEK, what everyone was discussing was certainly no longer the beginner's question of "can AI draw, write, or make videos?" That stage is long past. What is truly worth discussing is: when technology has become thoroughly tool-like, when more and more people hold this "brush," where exactly is the boundary of art, and where is the creator's place?
After sitting through the entire session, my biggest takeaway was actually not "will AI disrupt art?" The AI creation industry is moving from a "show-off phase" into a "division-of-labor restructuring phase."
Tools are getting stronger, generation is getting cheaper, and so a more essential question is being thrown back out:
When everyone can generate, what value is left for humans?
The Debate Is Not About Art, but About "Who Counts as an Author"
The most interesting part of this discussion was the disagreement among the panelists.
The first disagreement was whether "human intention" actually matters.
Liu Rushan's view was clear: what AI generates can certainly be art, but as long as this world is still controlled by humans, the boundary first lies in "intention." Without someone behind it pushing, selecting, and imbuing it with emotion and concept, content that purely runs automatically is hard to recognize as a genuine work.
Zhou Changyin's view was almost the opposite. He cared more about the result than the intention. As long as a work ultimately truly moves people, it is art; if what is generated is garbage, then it is not. Whether there is clear human intention behind it is less critical.
I think this exchange is particularly worth recording, because on the surface it is arguing about "art," but at its core it is arguing about something else:
Who has the right to be called an author?
One set of standards asks about attribution.
Another set asks about effect.
In the past, these two things were often bound together. Today, AI has forcibly pulled them apart. A work can "have effect" but not necessarily be recognized as having a clear author; it can also have creative intention but not necessarily truly move people.
Listening further to the on-stage remarks, you will find that everyone gradually pieced together a new sense of boundary: art is no longer defined primarily by "whether it was made by hand," but more by whether it contains human expression, human choice, and whether it can enter an interpretable, resonant social context.
Put simply, AI has not made the word art meaningless; it has merely shifted the threshold from "can you make it" to "what exactly are you trying to express."
The Bigger Anxiety: Will Creators Become "AI Designers"?
The second question at the roundtable was closer to the situation many practitioners face today.
When prompting, parameter tuning, card-drawing, fusion, and automation increasingly resemble a new production line, are today's creators acting as directors, or are they working for AI?
Xitang's answer was fairly direct, and very much like what a frontline practitioner would say: some people are helping AI do its work, and some people are making AI help them do theirs.
If you simply accept whatever the system spits out, you are of course more like an executor; but if you know what you want, know where the standard lies, know what to keep and what to delete, then you are using AI to amplify your own judgment.
Huang Naiyuan pushed this trend one step further. He said the future will become increasingly polarized: good creators will become more like directors, average creators will become more like designers, and these two groups will coexist for a long time.
I strongly agree with this judgment. Because AI does not necessarily elevate all creators evenly; it is more likely accelerating stratification.
In the past, many gaps could bebarely smoothed over with time, physical effort, and execution ability. Going forward, when execution costs are drastically reduced, what will trulywiden the distance between people will be judgment, narrative ability, and aesthetic density.
In other words, the more widespread generation ability becomes, the wider the gap between creators will be pulled.
This may be the real source of pressure in today's creation industry. It is not that AI replaces everyone in one fell swoop, but that it suddenly illuminates the layers that were previously hidden well enough.
When Results Get Cheaper, What Becomes More Expensive?
The question I most wanted to hear more rounds on throughout the session was the third one: will an efficient generation process cause art to lose its "soul"?
In traditional creation, we naturally believe that the process has value. You know how a film was made, what kind of struggle lay behind a painting, in what circumstances a work was born—these things in turn add weight to it.
But the problem with AI creation is that the process is compressed, even black-boxed. Many things happen inside models and GPUs, fast, cold, and clean, which makes one wonder: does something produced this quickly really still have a "soul"?
Liu Rushan's answercaptured a key point: precisely because results have become cheaper, the human process has become more precious.
It is not the few seconds of AI one-click generation that are precious, but the hesitation, tug-of-war, judgment, and expressive impulse that humans experience before and after. AI has no five senses, no suffering, no life experience of truly having lived once—these things remain the hardest-to-replace parts of creation.
I strongly agree with this. Because what it hits on is not "humans are forever noble," but "after generation depreciates, what begins to appreciate."
Previously, people placed more creative value on the output itself; going forward, it may increasingly be placed on how creators stuff their own experience, worldview, and trade-offs into the work.
Of course, someone on stage also reminded us not to romanticize "process." For many products, the reason process matters is not only because it is moving, but because it builds trust, letting users believe the result is controllable, compliant, and explainable. Plus, content itself has divisions: industrialized gratification content and works that hope to be taken seriously should not be measured by the same ruler in the first place.
This distinction is important.
Not all content needs a "soul narrative," but the more a work wants to enter the aesthetic and cultural judgment system, the more it cannot avoid human traces.
In the AI Era, What Is Humanity's Final Trump Card?
As the discussion reached its end, the question actually gradually contracted from "the boundary of art" to "humanity's trump card."
What can humans ultimately hold onto firmly?
Someone on stage mentioned personal stories, inspiration, and data assets; someone mentioned the judgment of "what is worth being seen"; someone also emphasized that decision-making power in open systems will remain primarily in human hands for the short term.
But what impressed me most was that very straightforward remark:
Where someone needs to take the blame, humans are irreplaceable.
Once a work enters the public space, produces impact, triggers misreading or controversy, someone has to come out and explain, take responsibility, and say "this is what I meant." Responsibility cannot be fully outsourced to machines, because the object society ultimately holds accountable is still a human.
At the end of the day, a creator's value is not only generation, not only expression, not even only aesthetics. More importantly, after the work is seen, are you willing to stand up for it?
This, in the short term, machines cannot replace.
The Boundary of Art Has Not Disappeared—it Has Simply Grown Back onto Humans
Returning to this Hangzhou roundtable, my biggest feeling is that AI has not made the boundary of art disappear; on the contrary, it has made it clear again.
In the past, a creator's scarcity may have lain in being better than others at drawing, writing, or editing. But going forward, these abilities will increasingly resemble infrastructure, more and more like water, electricity, and gas. What is truly scarce will instead become another set of things: why you want to express, what exactly you want to express, whether you have your own judgment, stories, and aesthetics, and whether, after the work enters the public space, you are willing to take responsibility for it.
AI has certainly given more people a pen.
But in the end, what determines whether you are a creator may still be that oldest question:
Do you actually have something you absolutely must say?
More Conversation Details
Unique Awards · Hangzhou AI WEEK Trends Roundtable Panel
"When AI Becomes the 'Brush': After Technology Becomes Tool-Like, Where Is the Boundary of Art?"
Guests:
MulanAI Founder & CEO — Liu Rushan
Vozo AI Founder & CEO — Zhou Changyin
Cutout.pro Co-founder — Cao Renyi (Roy)
Jifan Liuxing COO — Huang Naiyuan
Xiyangshi Founder — Xitang
Moderator: Unique Capital Partner — Wang Chaochao
Wang Chaochao: In the past, the brush was a physical object held in a painter's hand, dipped in pigment; but today? The brush has become code and models, dipped in data. The tool-ification of technology is already an undisputed fact, but it brings a rather ultimate question: after everyone holds this brush, where exactly is the boundary of art? Has it disappeared or been redefined? Shortly, please give your definitions. First, please each take one to two minutes to briefly introduce yourselves and what you are doing.
Liu Rushan: Hello everyone, I am Liu Rushan, founder and CEO of MulanAI. If the friends in the audience are the kind of deep AI core circle, you should know that MulanAI has already become a standard for judging whether you have entered the AI core circle. Because we almost never do marketing, but core-circle friends all know we were relatively early in doing workflows, and we have fully automatic Agents. We have a booth outside—just say your creative idea, and it will help you fully make a video that can be carefully modified. If you still don't know MulanAI, you may still be an AI outsider.
Zhou Changyin: I am Changyin, founder of Vozo AI. Before starting my business, I mainly did computer vision and graphics research at major overseas internet companies, and was fortunate to participate in some important global-facing products. After serial entrepreneurship, our team actually focuses on one question: why are there so many excellent video contents, but they can only live in one language? There are roughly 200-plus languages in the world, equivalent to 200-plus knives cutting the world into 200-plus pieces. Any language, video, or product—once it hits the wall of language, itsspread power declines. So we want to help every person, every video, every product cross these thresholds, penetrate into every market in the world, and help creators and marketers open up more markets. Actually, video translation is a form of secondary creation; how to make a work still move people after translation is a very artistic thing. It is a great honor to discuss with everyone how AI changes creation and dissemination.
Wang Chaochao: Simply put, it is using translation to break the dispersion and fragmentation caused by language, so that content can achieve the sameconvey meaning. Next, Mr. Renyi.
Cao Renyi: Hello everyone, I am Renyi, founder of Cutout.pro. Cutout.pro is already a relatively traditional AI company, founded in 2018. We mainly provide image and video APIs, with 20 million monthly views globally, an AI company that hastraverse both the small-model and large-model eras.
Huang Naiyuan: Hello everyone, I am Naiyuan, partner at Jifan Liuxing. We do 3D—using 3D to solve the hallucination problems that appear in 2D AI, providing more control andscrimp-able details. To put it bigger, we do world models; last year we open-sourced part of our world model code, which was cited by major companies' world models.
Xitang: Hello everyone, I am Xitang, founder of Xiyangshi. Our company benefited from the emergence of AI; we entered AI video in February 2024 when Sora came out. Our team mainly works in three directions: first, TVC advertising and short-drama production; second, we have self-developed tools; third, we also have a training business. Born on the same day as Sora, Xiyangshi is still here today.
Question One: Where Is the "Art Boundary" of AI-Generated Works?
Wang Chaochao: All five guests are people who make brushes, and also people who paint with brushes. Our first question: now when we see many AI-generated images, videos, and texts, they feel visually cool, but always feel like something is missing, just a little bit off. This raises a question: what is the most critical boundary for judging whether a work is art? Must it be drawn by a human, need physical traces, or is it enough that the work can evoke the audience's emotion? Mr. Rushan, please start.
Liu Rushan: I just specifically observed the expressive desire of theseveral guests behind me, and felt everyone is quite good at expressing themselves, so I will say less. I think whether what AI generates is art is already beyond doubt—it is art, just in a different form. So where is its boundary? I will only state the first boundary: intention (intent). Things generated unconsciously by AI—from a copyright perspective, if no one is operating behind it, no creative intention driven by emotion or concept—then what it produces has no copyright; but things we make out of intention have copyright. This is the first boundary of a work of art.
Zhou Changyin: I think art is just turning your feelings into a form that can move others. AI is just a means. I don't care whether you made it with AI or not—as long as the result can move people's hearts, it is art. Likewise, if what AI makes is garbage, then it is not art. I think the vast majority of AI-generated things today are garbage. But among them, some that truly move people—that is art.
Cao Renyi: On the topic of art, I think the definition just now was quite good: it can resonate with people. Recently everyone should have seen that popular video, the "snow mountain fox" (a video generated by OpenClaw). Every technological iteration, from traditional AI to current AI, has lowered the barrier for people to enter creation. When everyone can express their own ideas, such works emerge. Tools have not suppressed the emergence of art; they have promoted more people to participate.
Huang Naiyuan: I can use photography as an analogy. Everyone agrees that good photography is art, but you would not consider a casually snapped photo as art—it may just be your素材. So to judge whether AI is art: if good AI素材 can express your own intention, you know what you want, what you don't want, what can express your intention—then it is certainly art. If you just used AI to generate a seemingly exquisite image, then I think it is just素材. It still depends on whether there is emotion and meaning expressed.
Xitang: Whether a work of art itself was made by AI actually has no strong correlation—this is something everyone has first reached consensus on. The core lies in whether your work has expression, whether it incorporates your own uniqueness and storytelling. Now because the AI barrier is getting lower, after scrolling through short-drama rankings people have physiological reactions, because the homogenization is too severe. So the further we go, the more art leans toward personal uniqueness and storytelling—this is what is truly unique in the AI era.
Wang Chaochao: Let me summarize: defining the boundary of art under the background of extensive AI adoption—first, humans must have intention; second, the author must have human emotional judgment and choice-tugging in AI creation; third, placed in a social context it can produce resonance and controversy, and as a human you can explain and respond. These three points can form a new boundary.
Zhou Changyin: I actually have a different opinion. I think "intention," especially human intention, is not important. For example, if OpenClaw runs a work without human intention participating, but it moves people's hearts, then it is a work of art. The word "intention" itself is also something that cannot be precisely defined.
Liu Rushan: I quite disagree. I think a key point is that our society at this stage is still controlled by humans. Whether it is the copyright office or the courts, they first hold that AI cannot become an artist—an artist must be human. If a work is automatically generated by it, it simply will not be considered a work of art.
Question Two: Power Shift—Are Creators "Directors" or "Designers" Working for AI?
Wang Chaochao: Let us split the art discussion into human and object dimensions, and next talk about the issue of authorship. In the past, an author meant the person holding the pen, but in the AI workflow, a creator's work becomes prompting, parameter tuning, card-drawing, fusion. This brings identity anxiety: are today's creators directors with overall authority, or designers working for AI? Whose hands is creative power in? Xitang, please start.
Xitang: Some people may choose to be pure executors—whatever AI produces, I do—then they are helping AI do its work; while others make AI work for them. It is okay if you don't want to become a so-called AI director or AI screenwriter; this is actually very hard, and most people currently still work as card-drawers, animators, doing pure workflow execution. But what will be most precious in the future, beyond doubt, is whether you canharness AI well—whether you decide if what AI produces is good, your aesthetic standards—this is what is uniquely personal.
Huang Naiyuan: I think the conclusion is: good creators will become more and more like directors, average creators will become more and more like designers, showing polarization, and these two groups will coexist. For example, short dramas that cause direct gratification and dopamine secretion have a set of industrialized production pipeline standard logic; under this system you will become more and more like a designer. But if you want to create films like The Wandering Earth or Ne Zha, they require deep human creation, and such people will become more and more like directors. This is just like how there are both popcorn movies and art films in the market.
Cao Renyi: If you treat AI as a tool, like phone photography, then the effect a boyfriend takes and a photographer takes are certainly different. If you treat AI as a person, then AI acts as the cameraperson—it helps you execute the shooting, and you are the director.
Zhou Changyin: Content production has many roles: director, designer, and also producer (deciding what to make, what the qualification standard is). This is everyone's own choice—some people like to grind with AI every day doing design work, some like to be producers. Our video translation product is positioned to serve "producers": users just give the film to the software, and as for how to translate, how to handle emotion—these director-level fine tasks should be solved by the software. AI does not force humans into only one role.
Liu Rushan: I feel a little differently. With technological development, the "designer" role may disappear, and in the end all that remain are directors. For example, when MulanAI came out, its slogan was "your AI film studio"—you use an Agent to tell it what you want, and it basically helps you build it, no need for you to do detailed work. Why is OpenClaw so popular? Because for the first time people truly realized that saying one sentence gets someone eagerly working for you. In the future, humans only need to be directors, using willpower to make Agents work for them; designers may become a少数 genre like intangible cultural heritage.
Zhou Changyin: Actually logically, directors can also not exist. Just like Go—humans now completely cannot beat computers, but many people still play, although the meaning of pursuing the pinnacle of human intelligence has lost half. From a survival perspective, in the future editing really won't have much meaning, and even directors may not be needed.
Question Three: Will an Efficient Computation Process Cause Art to Lose Its "Soul"?
Wang Chaochao: Mr. Changyin has led us into the next question: which is more important, process or result? In traditional creation, the process itself has storytelling—for example, Van Gogh cutting off his ear, Wang Xizhi drunkenly writing the Preface to the Poems Collected from the Orchid Pavilion—the process endows the work with radiance. But in AI creation, code computes in GPUs, both cold and efficient. When pursuing ultimate effect, have we also lost the struggle and warmth in the creative process?
Liu Rushan: Because results can be obtained quickly, results have actually become cheap; under this premise, the process becomes more precious. But the precious process does not refer to the few seconds of AI one-click generation, but the process in which humans tug, hesitate, and think about how to better express emotion. This is our biggest difference from AI. AI has no five senses, has not experienced suffering. The first sentence of The Road Less Traveled opens with "Life is difficult"—every day's big and small sufferings are our nourishment; this is the point where humans and AI differ. When results become cheap, life's suffering and the creative tug-of-war process become extremely important.
Zhou Changyin: I recommend everyone watch a talk by Geoffrey Hinton (Nobel Prize winner in physics) titled "Will AI Surpass Human Intelligence?" When people think they are better than AI because we have feelings, self-awareness, and suffering, he believes these may be things that should be abandoned—humans sometimes think too highly of themselves. Returning to the question, I think for us doing B-end products, process is important because we need users to participate in the process to build "trust." Only through the process, letting them see that generation is correct and compliant, will they believe and accept the final result.
Cao Renyi: Let me give an example—Duchamp's artwork Fountain (a men's urinal). Why could it become a work of art? Because in this social context it caused discussion and emotional resonance, whether revulsion or debate—it is this "field" that made it. Its production process doesn't matter at all. For life, meaning lies in the process, but for a work of art, perhaps the result is more important; as long as the脑洞 is big enough, something produced in a few minutes is also art.
Huang Naiyuan: This depends on whether you are consuming "素材 (industrial product)" or "artwork." If I want to watch a domineering-CEO short drama to decompress, I don't care how it was made—I only care that the result gives me gratification; the result is most important. But if watching a film like The Wandering Earth, while the result looks good, if you trace its difficult shooting process and代入 yourself, you will feel the work is elevated. Just like buying a two-yuan bottle of mineral water you don't care about the production process, but buying Changbai Mountain snowmelt water, you will care about both the result and the process. It depends on the level of focus.
Xitang: Process and result are equally important in this era. If a creator is only doing repetitive work, the process is painful; but if the creative process (including your unique style, worldview, characters and other data) can, through continuous trial and error, reshape the creator itself, then this process is very enjoyable. At the same time, the result is also very important, because there are many excellent creators—"even good wine fears a deep alley"—good works need good results for marketing and presentation; one plus one can be greater than two.
Cao Renyi: One more point—just like our roundtable today, if everyone were just whispering in a corner, that itself would be a helpful process for participants; but we are sitting on stage being livestreamed and heard by more people, and this as a result has a more artistic presentation than whispering.
Question Four: In the AI Era, What Is the Final Trump Card Humans Will Firmly Hold?
Wang Chaochao: Since the wave of the technology brush is surging in, after AI becomes thoroughly tool-like, what is something that no matter how AI develops it cannot replace, and will surely be firmly held in the hands of human artists?
Xitang: I think there are two aspects: first, with AI's changes, itinstead helps shape virtual IP and idols, whose future ceiling and value are hard to imagine. Second, in this era, your personal stories, inspiration, and data assets. By effectively managing these data and collaborating with AI, sparking bigger creative ideas—this is precious in the AI era.
Huang Naiyuan: AI will increasingly know what you "like" (for example, recommendation algorithms get you hooked), but that does not mean it knows what is "worth" being seen by you. Dopamine content that simply satisfies human likes will be largely replaced by AI, but what humans need to control going forward is how to precisely express value content that triggers deep thinking and reflection. In terms of our business, we use 3D technology to control camera movement and scenes, precisely so that creators can more accurately express what they truly want to express.
Cao Renyi: At this stage, AI plays an excellent role in two areas: one is shaping new IP (for example, young Huo Qubing, Xin Qiji, etc., which can be presented without being restricted by capital); the other is extreme personalization (for example, home design), where AI can match everyone's different aesthetic will, bringing traditional culture and personalization to the extreme.
Zhou Changyin: Overall, it is certain that AI will keep eroding human territory. When AGI arrives we may not be able to hold our ground. But before that, what humans can grasp is the decision-making power of "open systems." If it is a closed system (for example, asking it to draw a picture, do specific translation), AI does very well. But open business decisions—such as whether to sell clothing to Brazil, whether to use translation software, how many Tokens to allocate to running OpenClaw—this kind of macro decision can only be made by humans.
Liu Rushan: There is a rather cheesy but very true saying circulating online: "Where someone needs to take the blame, humans are irreplaceable." It is the same in the art field. In a social context, when a work is exhibited and produces impact, someone needs to come out and take responsibility—explain the original intention, or apologize and pay for negative impact. This "taking responsibility" trump card cannot be replaced by machines in the short term, because power and responsibility are对等.
Wang Chaochao: Thank you to the five guests for your wonderful碰撞. Today we heard both the rationality of technology and the warmth of creators. As a tool, AI has not disrupted the essence of art, nor has it caused art to lose its soul; on the contrary, it has provided new possibilities for creation. It has lowered the threshold, giving more people the opportunity to express creativity, while also forcing us to rethink the ever-advancing boundary of art. The brush is technology, while creativity and emotion are the soul of art; in the AI era, perhaps everyone can have the chance to become "a modern-day Ma Liang with a magic brush." Thank you, guests, and thank you all!