[Kim Dae-sik Column] Art in the AI Era: We Must Not Forget That Humans Are Human

① Art in the Era of Mass Production of Imagination

Mass production of artworks is possible

The era of generative AI with emergent capabilities

Art for humans that contemplates humanity's eternal problems is a destiny

One hundred years ago, Europe was a world of shock and change. World War I destroyed what seemed eternal: Russia, Germany, Austria-Hungary, and even the Ottoman Empire. Many young people who returned from the battlefield having lost arms and legs or become blind said that the war was not what they had imagined. It was vastly different from the wars they read about in novels or saw depicted in statues scattered across European cities. No one could dodge bullets automatically fired from submachine guns, and tanks and airplanes decided victory or defeat. It was no longer individual bravery and skill but the state's mass production capability of machines that began to determine everything.

German philosopher Walter Benjamin believed that beyond economics and war, the mass production capability of machines would have a fatal impact on art and literature as well. If photographs and films could be reproduced thousands of times, why would we bother going to art galleries to see artworks? If highly sophisticated replicas indistinguishable from the 'originals' displayed in galleries could be mass-produced, wouldn't the difference between originals and copies become meaningless?

Perhaps we are living in a world where Benjamin's concerns from 100 years ago are rapidly becoming reality. Generative artificial intelligence (AI), which has learned from hundreds of billions of sentences uploaded voluntarily by humanity on the internet, has begun to possess language and cognitive abilities. Linguist Noam Chomsky calls generative AI like ChatGPT a "statistical parrot," but the more you converse with ChatGPT, the more unsettling it feels. How can we explain generative AI that shows emergent abilities to infer things it was neither taught nor explained?

Over the past 30 years, humanity has voluntarily uploaded sentences, images, and videos to the internet. Aren't all these ultimately human thoughts, imagination, and memories? If so, generative AI may not have simply learned sentences and images but may have absorbed humanity's dreams, fears, and thoughts. Is that why large language models (LLMs) now learn and predict not only simple sentences but also the interrelationships between text and images, images and videos, and videos and sounds? When you input "Draw Walter Benjamin talking with a robot," AI instantly creates a picture. Since the exact image did not exist on the internet, it is a truly new creation. And once a machine creates one piece, it can tirelessly produce 100, 1,000, or even a million pieces.

Image created with AI program (right) combined with Tolstoy portrait (Lee Young-woo, Art Team Manager 20wo@asiae.co.kr)

Literature is similar. The 19th-century Russian novelist Lev Tolstoy is famous for his epic 'War and Peace,' which spans 1,200 pages. Tolstoy wrote every word and sentence by hand and took six years to complete the work. But someday, won't AI be able to complete a 1,200-page novel not in six years but in six seconds? Then why should humans suffer and struggle for six years to create? Just as we no longer make plates and bowls by hand, shouldn't future artworks and novel adaptations for films be mass-produced by machines?

As we experience a world where imagination can be mass-produced for the first time in human history, we must seriously ask: Why should humans still create art? Why should humans still write poetry and novels and perform dance and singing on stage?

Mathematician Norbert Wiener imagined a future in his 1950 book 'The Human Use of Human Beings,' where highly automated machines would transform industry, economy, society, politics, and even art. Artist Nam June Paik, who became familiar with the machines Wiener called 'Cybernetics,' concluded that art for humans living in the cybernetic era is more important than cybernetic art itself.

Ultimately, the core of art in the generative AI era must be art for humans that contemplates the meaning of existence and humanity's eternal problems, beyond AI as a new tool in the era of mass production of imagination. Or perhaps the most important task of art is to remind us not to forget such questions and to remember that humans are still human.

Kim Dae-sik, Professor, Department of Electrical Engineering, KAIST

This content was produced with the assistance of AI translation services.