TIME COMMONS
时间公域
Copy Complete: Who Am I Now?
Style, Identity, and Authorship in the Age of AI
Give an AI a few dozen things you have written—emails, reports, social media posts—and it can learn your phrasing, your logic, even the way you persuade people. It can produce a version of “you” convincing enough to pass for the original: faster, cheaper, and never tired. This is already happening, and almost no profession can remain untouched by it. It cannot yet copy a whole person, but it can already copy the signals that make others say, “That is you.” An artist’s style is one of the most visible of those signals.
Moving between the Chinese and Western art worlds, I have repeatedly heard Western collectors, curators, and critics make the same judgment about Chinese artists—sculptors especially: the visual language is not unified enough; the artist has yet to settle into a style of their own. The implication is:
You still haven’t found yourself.
That sentence has shaped the professional formation of many Chinese contemporary artists: find your style, hold on to it, cultivate it. The more stable the style, the more recognizable it becomes; the more recognizable it is, the more easily it can enter the systems of galleries, collections, exhibitions, and auctions. Artists have been encouraged to compress the complexity of who they are into an identifiable visual language.
With the arrival of AI, this system has begun to turn against itself. The more successfully an artist turns themselves into a stable, recognizable style, the more readily that style can be converted into samples that can be learned, invoked, and copied. What once proved “this is you” can now detach itself from you and take on a life of its own.
And it is not only the language of painting that can be copied. AI is learning the phrasing of an email, the logic of a report, the way someone persuades a room. Artists are simply among the first to be forced to answer the question: if someone else can use your style to keep producing work that looks like “you,” once the copy is complete, who am I now?
Heteronym, Not a Pen Name
The Portuguese poet Fernando Pessoa (1888–1935), one of the most important Portuguese-language writers of the twentieth century, created dozens of “heteronyms” over the course of his life. These were not pen names in the ordinary sense, but authors with their own personalities, biographies, and systems of thought. Alberto Caeiro wrote of nature and instinct. Álvaro de Campos wrote of industry and sensory vertigo. Ricardo Reis wrote in a classical language of restraint and fate. Bernardo Soares wrote of the loneliness and metaphysical exhaustion of an urban office clerk. They differed not only in style: they commented on one another, and opposed one another on matters of politics and aesthetics. Pessoa gave many of them biographies, signatures, and astrological charts. Pessoa himself begins to look less like a single author than a vessel.
What this heteronym system did away with was not merely the requirement of stylistic unity. It discarded a deeper assumption: that one author must correspond to one stable personality. One person could become many authors. Those authors could be independent yet still shape one another. They did not have to justify themselves to one another, much less converge into a single voice.
A Genius’s Craft, an Ordinary Person’s Reality
Today, as AI begins to work through multi-agent systems—where different agents can have different roles, task boundaries, reasoning paths, and output styles while collaborating within the same system—something Pessoa could accomplish only through one person’s genius and a lifetime of effort has, for the first time, acquired an engineering structure that ordinary people can operate.
Agents and heteronyms remain far apart. Pessoa’s heteronyms possess literary histories, desires, contradictions, and bodies of writing developed over time; most agents are task roles. They offer one useful comparison: different positions of judgment can persist, divide labor, come into conflict, and leave a record of the process.
At the same time, the nature of style is changing. In the past, style was a capacity that took years of practice to form. Today it increasingly resembles an interface that can be learned, invoked, and transferred. Given enough samples, AI is already being used to identify elements of an artist’s visual grammar—brushwork, compositional logic, relationships of color—and then recombine them in continuing acts of generation.
Style has not disappeared. It is simply becoming easier to separate from its author and exist on its own.
“You Haven’t Found Yourself” Gets the Blame Wrong
“You haven’t found yourself” is one of the judgments I have heard most often over the years. Why does it sound so self-evidently true?
When we look back at Picasso today, we see the Blue Period, Cubism, Neoclassicism, and Surrealism, all arranged into a seemingly inevitable progression. But when Picasso turned toward Neoclassicism in the 1920s, many of his avant-garde peers regarded it as a retreat or a compromise. They were not looking at the Picasso of later art history. They were looking at a man who had suddenly changed direction.
Gerhard Richter offers another point of reference. He has worked in photorealism, squeegee abstraction, color charts, and glass installations—languages that can seem far removed from one another—yet few people doubt that they belong to the same author.
What actually does the work here is credit, not the number of styles an artist has—the point at which a name has accumulated enough credit for radically different languages to still be read as the work of one person. The same diversity can then receive two entirely different interpretations: in one artist it is called exploration; in another, failure to find oneself. The style has not changed. What has changed is the way others are prepared to read it.
That “credit” is not created by the artist alone. Galleries, curators, critics, collectors, markets, and art history collectively determine whose change will be read as exploration and whose will be dismissed as immaturity. An artist can explain why they changed languages, but they cannot single-handedly control whether anyone is willing to believe them.
What Chinese Tradition Saw Before the Age of AI
Chinese artistic tradition has rarely required one artist to correspond to a single form. Su Shi wrote poetry, practiced calligraphy, and painted bamboo. Xu Wei’s poetry, calligraphy, and painting moved between abandon and discipline. When Shitao proposed the idea of the “One Stroke,” he was not arguing for a single way of painting, but asking how the ten thousand methods could arise from one life.
Change was never the problem. When one life enters different media, it naturally leaves different languages behind. AI did not create this experience. It merely gives us a new way to describe it: one spirit can call upon many languages.
That proposition holds for Pessoa, for contemporary multi-agent systems, and for the Chinese artistic tradition. They come from different eras and different fields, yet they keep directing us toward the same question: why, here, did the language change?
Turning Pessoa into a Method
Artists have long worried about having their styles copied. Some fear that their work and language will be taken and published by someone else before either has had a chance to be seen, letting the copy reach the market first. AI has lowered the threshold for copying and made its source harder to trace. Pessoa did not answer this predicament, but he left behind a method worth reading anew.
His heteronyms were not an uncontrolled multiplication of personalities. They formed a carefully designed structure. Many had biographies, astrological charts, political positions, and relationships with one another. They wrote letters and commented on one another’s work, forming an authorial system whose history could be traced. Their stylistic differences were only the result. Why Caeiro wrote in one way and Campos in another was not concealed, nor left waiting to be resolved into a single voice. The “why” was itself part of the work.
Pessoa was not alone. Marcel Duchamp created a female alter ego, Rrose Sélavy. She had a name and a portrait made by Man Ray, and she became the credited author of some works. The art world understood her relationship to Duchamp, but that did not prevent her from being discussed as an authorial position in her own right.
Others have continued along this path in recent decades. Two artists make and exhibit work under the name of Claire Fontaine, a fictional artist whom they publicly describe as a “readymade artist.” Reena Spaulings is at once a fictional character, an artist, a gallery, and a collective identity; works made under the name have entered collections including that of the Museum of Modern Art in New York.
All these cases tell us the same thing: knowing that a figure has been constructed does not invalidate the work. Whether the name conceals one person or many was never the deciding factor. What matters is whether the judgments and languages gathered under that name can genuinely stand on their own, whether the labor is acknowledged, and who bears the consequences.
Multi-agent systems now offer another possibility. Different roles can persist within one system, each with its own tasks, tone, and reasoning paths, while discussing and correcting one another. An artist can use such a structure to build, test, and continually interrogate multiple “language-personas” under a single name. The reason for changing languages can enter the work itself instead of remaining locked inside the artist’s mind.
The Chinese idea that one spirit can call upon many languages does not need AI to prove it. Poetry, calligraphy, painting, and seal carving have long come together to constitute a person. AI simply gives that experience, for the first time, a working structure through which it can be clearly organized, recorded, and unfolded.
Style will continue to change, and it will continue to be learned. How, then, will we recognize an author? Consistent style will not be enough, nor will an explanation added after the fact. We will have to ask whether the reason for changing languages actually shaped the choices and the process—and whether the author is willing to bear the consequences.
This Question Does Not Belong Only to Artists
AI is learning far more than the language of painting. Given enough samples, it can already imitate a lawyer’s phrasing, a manager’s way of making decisions, or a teacher’s classroom rhythm—often more evenly than the person it is copying, without ever needing a break. Sculpture and installation are harder: spatial handling and material vocabulary do not break down as neatly as brushwork and composition. But parametric design and generative modeling are closing that gap, turning those languages into forms that can also be learned and called up.
Most people already live like systems of “multiple heteronyms”: one voice at work, another at home, and a third on social media. Yet few people treat this multiplicity as Pessoa did—as something worth designing and accounting for. Most simply hide it and carry it as an unspoken sense of fracture.
More identities are not necessarily better. Many people already carry too many roles. Designing our multiple “I”s also means deciding which roles can leave the stage, which demands we can refuse, and when we may choose to produce or perform nothing at all.
This question reaches everyone who moves between different settings and identities, not only artists. Collapsing ourselves into one unified self does not ring true; neither does expanding ourselves without limit. Can we account, as Pessoa did, for why we are this particular “I” at this moment—which of our “I”s are worth keeping, and which we can let go? That is the question we now have to answer.
“Who is the author?” was once a question confined to art history. Now it belongs to everyone who must live among multiple identities—and explain why.
