

If there isn’t a human being behind the work, the story it’s supposed to be telling ceases to be moving.
I n 1891, Mariano de Cavia, the most famous Spanish journalist of his time, published an article in the newspaper El Liberal chronicling a terrible event: A fire in the early hours of the morning had completely devoured the Prado Museum in Madrid. The article spared no detail, describing how the fire burned “on all four sides” and how several citizens were injured while trying to save the remains of works by Velázquez, Raphael, Rubens, Titian, and Goya. The chilling news spread throughout Madrid within hours, causing a huge sensation. But it was a lie. All the author wanted was to publish an experimental article that would draw the government’s attention to the museum’s poor maintenance and the risk of a major fire after two small fires had occurred in the preceding months — and he succeeded. The government remodeled the museum.
I just experienced my own “Prado Museum fire” — something experimental, and similarly illusory — and it has left me thoughtful. A month ago, I received a message from a friend, a Spanish musician. He sent me a Spotify link to a young artist who had just released her first album. He told me he was her producer. The music sounded relaxing, and the girl’s voice was perfect — whispering, full of character, delicate. The album was the perfect accompaniment to a bath or a glass of wine at home in dim light on a rainy afternoon. The girl happened to be very beautiful.
I congratulated my friend on the release and asked him to arrange an interview for my music podcast. “It’s not possible,” he said. “Why?” I asked, surprised. After all, whenever someone sends me an album, it’s usually so I can write a review for my music magazine or interview the artist. “Because the girl doesn’t exist,” he replied. “What do you mean she doesn’t exist?” I asked, bewildered. “It’s all my own invention, created with AI. But it’s a secret,” he finally replied.
Although I love the analog world and am a hopeless nostalgic, I’m no Luddite, nor do I live apart from new technologies. My music magazine produces a weekly television program using the latest technologies, including virtual studios and all kinds of AI programs. Purely for fun, I’ve experimented with everything artificial intelligence can do. When I published my last novel, we launched a special website where the different characters in the plot came to life and showed their faces, as AI applications imagined them based on the original text.
I’ve also worked part-time in music journalism for 25 years. I have a certain idea of what’s good or bad and what I like or don’t like. Dozens of artists ask me for help or guidance when they’re composing songs. I had a band for almost a decade, and my life today wouldn’t be the same without Spotify constantly playing.
And yet my friend managed to fool me in two minutes with a damn AI-generated song.
This had a profound impact on me. As soon as I learned that the girl didn’t exist, her voice lost all appeal. And as soon as I learned that the melodies were created by AI based on settings chosen by my friend, they ceased to be moving and relaxing and instead became unsettling and disturbing. Of everything on the album, perhaps the only truly human element was my friend’s lyrics, but even those had to be adjusted by AI to sound natural alongside the music. I felt a mixture of surprise and disgust, as well as deep shame at having been deceived.
Music created by a machine has a sinister and dehumanizing quality. AI-generated artwork resembles the laughter of a madman or the deranged dream of a villain in a science-fiction film. AI-generated poetry is empty, bland babble. AI-generated artistic photographs are confusing, unsettling, and disgustingly perfect — except when they contain six-fingered humans or garbled signs. After my friend’s revelation, my conclusion was almost immediate: If there isn’t a human being behind a work, the story it’s supposed to be telling ceases to be moving. If there is no soul, there is no art.
I’ve finally realized that I’m part of the 97 percent: the share of people unable to distinguish a human song from an AI-generated one, according to a Deezer survey.
So far, most debates about art and AI have focused on legal issues, the damage to cultural industries, or the risk that artists will start copying AI, thereby plunging their talent into a well of stifled creativity. I miss the human element in this debate. Birds that build elaborate nests do not create works of art, just as moles that create masterful underground tunnels are not architects designing transportation systems. Animals cannot be artists. Art is either human or it is not.
In 2017, the Polish poet Adam Zagajewski delivered a magnificent speech in defense of poetry after receiving the Princess of Asturias Award for Literature. After admitting that we live in times that are “not conducive to poetry,” he explained that we experience a savage duality between the pragmatic and that which belongs to the realm of imagination or beauty. “Poetry is not in fashion; crime fiction, the biographies of tyrants, American movies, and British television series are in fashion,” he said. “Politics is in fashion. Fashion is in fashion. Relationships are in fashion. Substance is not in fashion.” Zagajewski defended the importance of both sides of this dualism: “We cannot forget evil, injustice which continually changes shape, the things that perish, but neither can we forget happiness, the ecstatic experiences that the thick handbooks of political theory or sociology have not managed to predict.”
Art was born to give meaning to the world and to ourselves — and to transcend reality itself. The history of art is the history of humankind: of our fears, our loves, our search for meaning, our passion for beauty, our striving for perfection, our attempts to unleash the torrent of creativity within us, and our need to connect emotionally with others. Because no AI can achieve any of these aspirations, we should not call something created by a machine a “work of art.”
Furthermore, one of the central premises of artistic and intellectual life is that every new work of art is precisely that. Leaving aside major historical examples of plagiarism, no work of art that copies another is considered new. We accept that every creation is infused with a mysterious mix of muses, influences, references, labor, and personal talent. However, AI does not create genuine works of art. These programs are trained on millions upon millions of previous creations across every artistic discipline. They will never create anything entirely new, nor can they ever be truly avant-garde, because their very nature prevents them from breaking free of what already exists or embracing something truly groundbreaking and authentic.
Moreover, AI cannot make mistakes in the way that humans do. Error — exemplified by the torn canvas or the discarded song — is an essential part of the work of those who aspire to be great artists. No musician releases the first song that comes to mind; a song is the fruit of thousands of decisions and experiences.
Music journalists often say that a particular album or song is special because the artist created it “from the heart.” It is obvious when a band mechanically composes a hit that simply aims for No. 1. It is also obvious when a hit song is born from an intimate and real experience. AI cannot have experiences of this kind because it cannot fall in love, break up with a girlfriend, suffer the fear of war, or value the embrace of a friend.
Artists are concerned about the rise of AI and the hypothetical risk that it will end up taking over their work. Many musicians tell me they fear that composing and recording albums will become meaningless, or that AI-generated music will become fashionable among artists, contaminating everything with its lack of soul and leaving the industry complicit in deceiving listeners. But understandably, artists are most concerned with the economics of AI. First, because they fear losing their jobs. And second, because they believe they deserve royalties from AI companies, whose applications have been trained on their work without their knowledge or consent. This last point is the subject of a long-standing legal dispute in Europe that will soon be resolved once a consensus is reached. The European Union has for months urged member states to find ways to compensate artists who feel they are being exploited by AI systems that extensively consult their work.
Mechanisms for separating human and AI-generated creations across the various platforms artists use to reach wide audiences are becoming increasingly advanced. Platforms like Spotify are closely studying the phenomenon of AI-generated artists, which are already generating significant benefits for their creators and amassing record-breaking streaming numbers. I imagine that all these debates will either be resolved or evolve as AI technologies are perfected, legal frameworks regarding ownership and rights are refined, and new tools emerge. However, the most important debate cannot be postponed: We must urgently reclaim the human element in artistic creation.
In our postmodern world, artistic nihilism sometimes makes it difficult to find the human element in every work, regardless of artistic discipline. But if we view the history of art as a vast continuum, it becomes clear that the human element is the sine qua non of artistic creation. Perhaps it is time to frame the debate in these terms: AI cannot create art; it is merely an illusion. Even if AI eventually manages to create technically perfect songs, paintings admired around the world, and poems that can move us to tears, it will remain nothing more than empty noise. Perfection is not a reason to consider what AI produces art, because perfection is not a typically human quality. Imperfection is central to all artistic creation, and even if an AI can one day imitate it, the result will still be nothing more than ones and zeros arranged by sophisticated programming. It is a Chestertonian idea: machines manufacture identical objects, but humans — the artisans — introduce variation, surprise, and their own stamp of individuality.
More than a hundred years ago, Leo Tolstoy seemed to anticipate this dilemma of generative AI and tried to settle the matter once and for all. He sought to preserve the art of the future from contaminations and threats. “Every work of art establishes a relationship between the person to whom it is addressed and the one who produced it, and with all those who, simultaneously, before, or after, receive an impression of it,” Tolstoy wrote in What Is Art? “The word that transmits the thoughts of men is a bond between them; the same is true of art. What distinguishes it from the word is that the latter serves man to transmit his thoughts to others, while, through art, he only transmits his feelings and emotions.”
A world without art, or a world with non-human art, is something much closer to hell: dehumanized, cold, and more dangerous for mankind.