Is art art if there's no artist?
The debate about artwork and artificial intelligence (AI) was rekindled recently after a painting by a computer won the top prize at an art exhibition in the United States last month.
Those claiming that pieces by programs are "products" that plagiarize existing images and those arguing that they are "artwork" that add creativity by using technology are clashing. Computers have been used to create visual artwork, compose music and write poetry and novels.
At the Colorado State Fair fine arts competition last month, a man named Jason Allen took the first place in the digital art category. "Théâtre D’opéra Spatial," the work submitted by Allen, was created with Midjourney. The computer program is able to create four images within five minutes when a user enters a message in the chat window. The user can then select a specific image and change the color and shape, enlarge or reduce it, and change the background and style.
Midjourney is an example of multimodal AI on which global big tech companies are focusing. Multimodal AI is an AI model that accepts and processes various bio signals, such as text, images, voice, motion and facial expressions.
Big tech companies have been concentrating on the development of super-giant AI that has significantly expanded information processing functions compared to existing AI. A supergiant AI is an AI that has been trained to learn a lot of data from language information and makes human-like judgments based on it.
A multimodal AI goes one step further and accepts and responds to bio signals of various aspects — or modalities — in various ways. Because it also accepts and processes information other than language, a multimodal AI can produce results similar to human creations. This means that AIs may soon exceed the capacity of the human brain.
Tesla CEO Elon Musk said in an interview with the New York Times in 2020 that "we're headed toward a situation where AI is vastly smarter than humans, and I think that time frame is less than five years from now." Musk himself is focusing on developing multimodal AIs through OpenAI, an company founded jointly with Sam Altman, founder and former president of Mountain View, California's Y Combinator.
OpenAI introduced Dali last year, an AI painter that visualizes text. Dali 2, which was released earlier this year, is said to have improved originality and artistry compared to the previous version. Google's Imagine, Nvidia's GauGAN2, and Kakao Brain's Karlo also create images based on text.
Each AI artist has a different style of painting. The Next Rembrandt, created by Microsoft, reproduces the style of Dutch painter to create portraits in the way that he might have painted.
The first poetical play written by AI, titled "Paphos," was staged at the Arko & Daehakro Arts Theater in Seoul last month. The author of the play is SIA, an AI poet created by Kakao Brain. SIA learned how to write poems by reading about 12,000 modern and contemporary Korean poems. If a user inputs a keyword and a command, SIA composes a poem in one second.
"AI will eventually develop into a form that makes humans more human and asks fundamental questions about art," said Kim Je-min, the professor of theatre at Seoul Institute of the Arts who directed Paphos.
Last year, a full-length novel written by Birumpung, an AI developed by the Graduate School of Artificial Intelligence at Pohang University of Science and Technology, was published. Using input information, the story unraveled over 500 pages.
Music composed by AI is already being used in various ways in movies, television series and advertisements. In Korea, POSAlabs, Upvote Entertainment, CreativeMind and Neutune are presenting music made through AI. EvoM, an AI composer developed by the Gwangju Institute of Science and Technology in collaboration with CreativeMind, composes a variety of music ranging from classical music to trot.
The logo song for the web drama "Gaus Electronics" was composed by an AI developed by Genie Music and Upvote Entertainment.
Tilda, an AI character developed by LG AI Research, is equipped with the multimodal AI EXAONE as its brain. Tilda, together with designer Park Yoon-hee, created a costume around the concept "flowers from Venus" and presented it on a stage at New York Fashion Week earlier this year. The goal of technology companies jumping into the development of multimodal AI is to make AIs that can replace all human activities. It is expected that multimodal AI will be available in various service industries first, such as education and product consultation.
Changes in social perceptions and systems are slower than the pace of technological development. As with "Théâtre D’opéra Spatial," the debate over whether AI-made works will be recognized as creations is still ongoing. Many countries, including Korea, regard creative works that express human thought or emotions as artwork.
An ethical standard for AI that resembles humans is also needed. Kakao recently established a technical ethics committee for the first time among domestic companies.
"The system as of now is insufficient compared to the pace of AI-related industry and technology development," said Son Seung-woo, a professor of industrial security at Chung-Ang University. "It is urgent to discuss the protection of creative works and legal and ethical guidelines."
BY KIM KYUNG-MI [email@example.com]