Source: Sichuan Daily
In recent years, the emergence and rapid iteration of artificial intelligence (AI) has attracted increasing attention, and has formed a huge impetus and impact on the global cultural content production, artistic creation, and design industries.
In late May, the organizing committee of the 2023 Lishui Photography Festival officially announced that it will set up an AI video art award for the first time since the establishment of the 2004 photography festival, calling for video art works created or assisted by AI technology, requiring contributors to provide original pictures, prompts and relevant explanations to introduce the application and creation process of AI technology in works. In addition, this year's festival will also hold a forum on AI images, inviting all parties to engage in dialogue and exchange on related topics.
Almost at the same time, the official Chinese version of the AI painting tool Midjourney began to be closed beta, as long as you enter the text you think of, you can produce the corresponding picture through artificial intelligence, which takes only about 1 minute; Similarly, Adobe Firefly, released in March, expands the original image based on text.
The influence of artificial intelligence on art has gradually penetrated into the fields of fine art, calligraphy, photography, and design, including animation, games, clothing, jewelry, etc. Different people may have different views on AI art, but there is no doubt that changes have taken place and will continue to shake the current art ecology and industrial landscape.
traceability
Not "suddenly"
Artificial intelligence has come to life
At the undergraduate and master's levels, Li Mao is a pure "art student" specializing in oil painting. Later, at the suggestion of his mentor Cheng Cong, he was admitted to the School of Computer Science of Sichuan University, where he studied under Professor Lv Jiancheng, the dean, and engaged in research in the field of computer generative art. Today, Li Mao is an associate professor at the college.
According to him, in the fifties and sixties of the last century, some scientists and artists gradually noticed the artistic creation potential of computers. However, the narrow exploration of artificial intelligence painting is generally believed to have begun at the beginning of this century, and the early stage is the "style transfer" stage. Specifically, it is to provide a stylistic image (such as Van Gogh's impressionist painting), convert any image (such as a city street view photo) into that style, and preserve the content of the original image as much as possible.
The subsequent "generate graphics" stage, that is, directly converting text into images, once encountered technical bottlenecks: first, the generated image resolution was low and a lot of details were lost; Second, it is not stable enough, just adjusting the order of the text, you will also get completely different results. It was not until the emergence of the "diffusion model" at the end of 2021 that "low-resolution" text could finally be accurately converted into "high-resolution" images, and AI painting became scientifically mature.
Also paying attention to this field earlier is Chen Wei, a well-known digital artist and illustrator in our province. Around 2000, Chen Wei, who has a science background, began to explore the art of digital illustration, and is currently making great achievements in creative practice and theoretical research. "Anime and technological development are highly bound." He said that some older versions of AI painting tools produce images that are sometimes jokingly referred to as "stitching monsters" and cannot be commercialized; At present, the more mature AI models on the market can trigger the working mechanism through keywords to quickly complete a work, "although it may not be perfect." ”
Perhaps many people feel that the wave of AI is "sudden", and it seems that overnight, ChatGPT conversations, Midjourney images, "AI Sun Yanzi" and so on filled the online world. However, some experts and scholars point out that in our current art practice and even in daily life, the influence of AI has long been everywhere.
For example, the well-known "night mode" of mobile photography usually requires holding the hand steady for a few seconds after pressing the capture button, but this has nothing to do with the logic of the camera "extending the exposure time to improve sharpness". In just a few seconds, the phone took a large number of photos, and then synthesized and output a clear photo through the AI model. Therefore, the photos of "night mode" are not only "taken", but also "calculated" by AI models.
challenge
Good at "cross-border integration"
The impact extends far beyond the realm of art
Chen Wei said that AI will break the professional division of the industrial era and will bring a new round of technological revolution and greatly change people's concepts. "In the past, we used computers and mobile phones to collect information, find pictures, and then integrate them ourselves; But AI as a 'cross-border integration tool' replaces this part of the work, we just have to make a choice. ”
"AI is currently just a tool, but unlike traditional tools, it grows and iterates extremely quickly." Chen Wei found that he had not yet used it deeply, but he already had a sense of dependence on AI.
Naturally, this also leads to some new problems, causing anxiety for many people. Chen Wei summarized it in two aspects: AI, as an integration tool, will undoubtedly bring a lot of copyright disputes; The powerful ability of AI has made many creators feel entangled, and some young people even feel that the future is hopeless, full of frustration and crisis.
This concern is not unfounded. Cheng Zhuo, a discipline leader and professor at the School of Photography of Sichuan University of Communication and Communication, said that tradition believes that art is based on certain techniques, and the Latin etymology of English art (art) has the meaning of "craftsmanship and skill", and those who reach the state of pure youth are artists. With the intervention of AI, "technique" may no longer be an important criterion for examining art, and the boundaries of art will be greatly expanded and dissolved at the same time.
In the field of photography, the most obvious impact of AI is commercial photography. "For example, the clothing season needs to be produced in batches, and after AI intervention, there is no need for photographers, models, and venues, and the post-effects can be rendered arbitrarily through keywords." Cheng Zhuo found that there are already institutions that use the "information gap" brought by AI to make profits.
In anime and game original paintings, such as "two-dimensional" paintings, because there are too many samples that can be provided to AI for training, they are quickly broken. For enterprises, reducing costs with AI is an irresistible temptation. Chen Wei made an estimate: a game company, a project team of 100 people, is expected to save 70%-80% of labor costs. At a further extreme, if all of Jet Li's martial arts moves are used to train AI, there may be no need for live actors.
In the eyes of some experts and scholars, "literature ethics" will also be a major challenge brought by AI. Cheng Zhuo said that before the birth of photography, it was the history of words and painting, and after the birth of photography, "one picture is worth a thousand words", establishing the history of image vision and changing people's way of understanding history. However, the picture of AI "out of nothing" may be difficult to distinguish with the naked eye alone in the future, will this confuse history? If photography provides evidence for history, AI may make "authenticity" itself problematic, with implications that may extend far beyond the realm of art.
foreground
"Light technology, heavy thinking"
Find new opportunities in the "labor pain"
"Compared with the breakthrough of AI, human thinking is limited and easy to solidify." Li Mao believes that these factors may promote the cheapness of art works, "When we have a lot of images, the value of images will decrease." In his opinion, painters and photographers below the middle level may be robbed of their jobs by AI.
But that's only one side of the coin. On the other hand, with the continuous development and improvement of AI, it may also provide new breakthroughs for the development of the industry and even give birth to new jobs. Chen Wei's understanding of this is that AI has replaced some jobs, but opportunities have also come, just saying that "opportunities may be beyond common sense."
Taking digital illustration as an example, Chen Wei believes that after the emergence and widespread application of AI, there will be many "amateur" illustrators with non-professional backgrounds, and even theoretically everyone can use AI to create illustrations. In this case, an author who is sensitive to text can complete excellent picture book works through AI; Conversely, an illustrator can also use AI to write novels and scripts with a strong "sense of picture" to achieve crossover.
Cheng Zhuo also believes that AI brings "labor pain" and may also give birth to new opportunities. "After the invention of photography, many portraitists lost their jobs, and they were forced to change careers to learn photography and become photographers. Some of the early important figures in the history of photography, in addition to physicists and chemists, were also painters. For example, the Frenchman Louis Daguerre, a former stage background painter, invented daguerreotype photography in 1839, which is also regarded as a symbol of the birth of photography.
AI may also spawn branching industries, such as model training, big data analysis, and retouching. "The accuracy of AI images is not yet high, and the early creative planning and post-extension services also need to be improved." Chen Wei believes that AI brings not a problem of unemployment, but a problem of transformation, that is, letting people "put themselves in their place".
If we put aside the issue of industry and employment, experts and scholars generally believe that there is no need to be too anxious about the impact of AI on artistic creation. The current AI art can only be said to be rich and full of details, but as Li Mao said, artistic creation is diverse, not all realistic, than details, so there is a lack of quantitative indicators for the level of judgment.
Cheng Zhuo believes that from the artistic level, it is still difficult to judge the impact of AI, but there is no doubt that "artists have new tools" and will definitely use it to create. Chen Wei said that artistic creation should be "light on technology and heavy on ideas", and if you only want to be a "painter" and only pay attention to technology, you will soon be replaced like the textile workers after the emergence of the Jenny spinning machine; To be an "artist", use different media and technologies as tools to express ideas. (Sichuan Daily All Media Reporter Yu Rubo)