Li Houqiang
(Deputy Director of the Decision Advisory Committee of the Sichuan Provincial Committee of the Communist Party of China and the People's Government of Sichuan Province, Chairman of the Chengdu Municipal Federation of Social Sciences, Second-level Professor and Doctoral Supervisor of the Sichuan Academy of Social Sciences)
The imagination of artificial intelligence (AI) is to walk between wide and narrow. The impact of AI on literature and art has infinite possibilities and physical limitations. The limits of the chip, the size of quantum, the depth of the universe, the complexity of the human body, the subtlety of emotions, the explosion of inspiration, the enlightenment, the interruption of skills...., there are a series of paradoxes and unsolvable. AI art can free up thoughts, but it must be accurately landed. Human cognition is constantly advancing. It is not possible for AI to be anti-servant.
The first concerns the relationship between art and science
Regarding the relationship between art and science, the 19th-century French writer Flaubert once said: "Art is becoming more and more scientific, science is becoming more and more artistic, and the two break up in the foothills, and one day, they will meet again on the top of the mountain." Li Zhengdao, a famous Chinese physicist and Nobel laureate in physics, also said: "Art and science are two sides of the same coin, they originate from the noblest part of human activities, and both pursue profoundness, universality, and eternity." ”
What is art? Art is man's telepathic and creative expression of natural and social beauty. What is beauty? Beauty = 96% symmetry + 4% breakdown, beauty = 85% objectivity + 15% subjectivity. All natural beauty can be expressed in terms of frequency (reciprocal of wavelength), including visual beauty, auditory beauty, olfactory beauty, taste beauty, and tactile beauty, because matter is energy, is a wave string. For example, the color, red-orange-yellow-green-cyan-blue-violet, is different in wavelength. Music is a combination of sounds of different wavelengths. Social beauty such as spiritual beauty, language beauty, and behavioral beauty cannot be expressed in frequency at present. Art is the creation of people and the pearl of culture.
Artificial intelligence (AI) is a new technical science that studies and develops theories, methods, technologies and applications for simulating, extending and expanding human intelligence, which can simulate human consciousness and thinking processes. Artificial intelligence is actually a very good tool that can effectively help people learn, including process guidance, result judgment and content analysis, which is also an important function of this technology.
Since the birth of artificial intelligence, the relevant theories and technologies have become increasingly mature, and the application fields have been expanding, and naturally and inevitably gradually spread to the field of art, many artists try to use this technology in creation to innovate content and form, and thus artificial intelligence art (also called neural network art, is a new form of using algorithms to create).
Art is a uniquely human expression of ability to embrace beauty and create beauty. Through the works, people experience the artist's experience and thinking, and receive expressions and emotions beyond the object. Of course, the beauty in it comes not simply from form, but more from some kind of behavior, logic or even judgment. People will continue to generate new thinking and judgments in the future, and will also make art focus on more different fields, from nature to society, broad and vivid. If there is anything in the future that will make people human and open the gap with artificial intelligence, this answer is art. In other words, in the artist's mind, those who can create and appreciate beauty are the ones who are truly human, and those who will not be replaced by technology.
The greatest value of digital technology for art and design creation may be the new art design ideas, forms and models that emerge after using artificial intelligence as a co-creation partner.
AI art is a collective term for the art created by the computer automatically, from the 50s to the 60s of the 20th century, people gradually saw the potential of artistic creation on the computer, and now AI art has become a new and important form of art.
The first to explore computer painting were not artists, but computer experts and engineers, and oscilloscopes were the main form of early computer painting. By the mid-60s of the 20th century, computer-generated art was accepted as a new form of art, and more computer experts and artists entered the field. In the 21st century, deep neural networks have promoted the rapid development of intelligent painting.
AI art began to be commercialized in the 60s of the 20th century, gradually being collected by institutions and individuals, and it was only in the late 90s of the 20th century that it was included in the museum's exhibition and support programs.
The recognition of artificial intelligence painting is particularly prominent in the field of collecting. At Christie's auction in October 2018, a painting caught the eye, with blurred silhouettes, psychedelic demeanor, and full of artistic conception, which at first glance seemed to be the work of a famous artist. In the end, the AI painting "Portrait of Edmond Bellamy" was sold for $432,500, which Christie's said "marks the arrival of AI art on the world auction stage".
AI art is currently in the stage of expressing human innovation, ideas and emotions with the help of computer tools. The most essential feature of AI art in the true sense is that the creativity, ideas and emotions of art are partly or completely generated by AI, that is, AI has "inspiration".
Second, about the impact of AI technology on art
AI technology can help artists better perceive things, understand human nature and culture, and provide more innovative and subversive ideas and means. At the same time, the application of artificial intelligence technology can also reduce the time and cost of artistic creation, and promote the transformation of art in creation and communication.
When artificial intelligence and human intelligence collide, it is bound to inspire a new round of revolution in art design. For example, artificial intelligence and art designers have different interpretations of the same phenomenon, different answers to the same question, different strategies for the same goal, different interpretations of the same theme, etc., the fusion of the two wisdoms can evolve endless works far beyond expectations. The impact of AI art includes: 1. In the field of digital art creation, AI subverts visual art. 2. The combination of AI and graphic design not only lowers the threshold of design aesthetics, but also brings a more inclusive and efficient form of creation, which is recognized and supported by most people. 3. AI painting programs will inevitably hit the low-end production market of some functional paintings.
Advantages of AI art: 1. Improve the productivity of the art industry; 2. There is huge commercial value; 3. Change the way illustrators and designers work; 4. Mass production of beautiful images.
Shortcomings of AI art: 1. AI currently has no way to change feedback according to customized needs; 2. AI mapping currently only focuses on imagination and lacks logic; 3. Whether the final effect of an AI mapping is good or not, with a certain element of luck.
In short, AI has the emotional limit of creative motivation, the imagination limit of artistic expression and the value limit of the effect of the work in terms of artistic creation, and the lack of consciousness and thought is its fundamental limitation. The new achievements of the combination of AI and art have many deficiencies in the form of expression and quality level, and many related works of art have only established a relatively superficial and superficial relationship between AI and art, reflecting that the integration effect of AI and art is not ideal. What AI analyzes may not be the heuristics of the artwork, but the type of structure it has. Combination does not equal innovation. AI doesn't generate any inspiration, it just combines foreign ideas in a simple way, and it doesn't contribute anything to the generation of ideas. Because it is impossible to solve the problem of connecting heuristics and structures and the problem of AI's artistic personality, AI that independently generates heuristics still cannot create real works of art.
Third, new possibilities for AI art
Art is always an activity "for people", and without people's needs, people's feelings, people's minds, and people's activities, art has no meaning, or art does not exist. Throughout history, human technology has been developing and constantly promoting the development of human art and design, but it has never subverted the basis of "being a person" in art design, and digital technology is the same. The emergence of artificial intelligence and its beginning to play a dominant role in art design may shake the convention of art "by people" to some extent, but it still has to act on people (perception, appreciation, evaluation, etc.) to become art, and from this point of view, people are always an indispensable element of art.
The computer-generated creative pattern can realize the uniqueness and uniqueness of the public pursuit to a certain extent, meet the pursuit of personalized customers and realize the personalization of industrial products.
New possible aspects of AI art include:
1. AI is used in the generation of Chinese brushes, freehand, and abstract paintings, including following the way artists use sketches and reference pictures to create, and using computer technology to understand the color and mathematical relationship of reference images.
2. Use AI to train the beauty rules of form, such as the balance, density, rhythm, rhythm, golden ratio, rule of thirds, etc. of the visual elements of the picture, to form a generative model of abstract painting, and use the purest points, lines, colors, textures and other elements as the expression elements of abstract painting for creation.
3. The machine learning model using AI, this model is not simply a bunch of pictures stitched together, but on the basis of keywords, the image is iteratively rendered again and again, until the machine determines that the rendering result matches the given keyword the most, and the rendering result will be finalized.
Fourth, about overcoming "AI psychological phobia"
For AI technology, we cannot blindly worship, let alone worry and fear. AI is essentially a human tool, a technical means, not a creative subject, not a poet and artist, but can stimulate the creative potential and enthusiasm of writers and artists, and become a pacemaker of inspiration and a source of power. At present, AI is an extrapolation of knowledge and ability for individuals, but for the entire human race, the knowledge and ability presented by current AI are still internal interpolation, but more dense.
Massive data, advanced algorithms, improved computing power, deep learning, reinforcement learning, and generative transformation are the main features of ChatGPT. This is the generative pre-trained transformer GPT (Generative Pre-trained Transformer), which can understand and process human text language and natural language, not a simple chatbot. Compared with previous analytical AI, ChatGPT realizes generative AI, which learns to summarize the distribution of data while learning the patterns generated by the data, and uses autoencoders to create new samples that do not exist in the data. The essence of AI art is "database creation", which is the learning, imitation and combination of human works. At present, the poetry, calligraphy, oil painting, music and other works of art that have been created have a strong sense of patchwork, strong randomness, no personal style and technique, and appear very naïve, and are industrial products rather than works of art. It is difficult for ordinary people to distinguish between AI works and handmade works, but real experts know the difference at a glance. For the creation of novels and large-scale essays, it is just a simple accumulation of multiple materials and elements, and it is difficult to see excellent works. Emotions and passion, beauty and imagination, understanding and inspiration are difficult obstacles for AI to overcome. Without your own social experience and psychological journey, it is difficult to write big works with personality. However, with the improvement, enhancement and perfection of algorithms and computing power, new forms of AI may appear, even close to human intelligence, emotions and emotions, and full perception capabilities, AI art will have new breakthroughs, and even replace some handmade works. AI art also involves copyright issues, the ownership of products, which goes beyond traditional literary and artistic creation, and cannot be clearly defined and judged at present, and algorithm designers and data collectors should be studied in the future, who may be the real source of creation, using the "rights sharing" scheme. In short, robots can completely transcend individuals, but not the overall pattern. AI art is dominated and controlled by people, and there is a limit state. Cognition is infinite, but physical nature has limits.
One is the chip (transistor) limit (channel length). The theoretical limit of a chip is 0.2 nanometers in diameter for a silicon atom, with both thermodynamic (temperature) and quantum mechanical limitations (tunneling). Studies have shown that each parameter such as integration density, clock frequency, supply voltage, and number of wiring layers limits the overall performance of the chip. Theoretically, the limit size of a single device is 1 nanometer. However, when billions or billions of transistors are integrated into one, there are more constraints such as heat generation, wiring and so on, so chips have limits. As understanding deepens, there may be new breakthroughs.
The second is the quantum limit (Planck's constant). Quanta has sizes, and Planck length is roughly equal to 1.6x10-35 meters, or 1.6x10-33 centimeters, which is one-tenth the size of a proton. At a small scale, it is difficult to make new breakthroughs. As awareness advances, there may be new directions and approaches.
Third, the complexity of human beings cannot be completely and completely imitated by machines. The complexity of the human body, especially the neural network system of the brain, is similar to the universe, and there are many unknowns (cells, genes, blood vessels, meridians, acupuncture points, neuronal links, etc.), which AI cannot fully learn, simulate and imitate. The human body is far more complex than any machine! AI has a long way to go!
Fourth, there is a series of social and natural paradoxes. For example, is the robot a human or a machine? What is its life? What is its DNA? Where is its reproductive capacity? Where does its inspiration and emotion come from? There are also social philosophical issues, such as the difference between technology and literature and art, the difference between algorithms and chips, the difference between desire and reality, the difference between inspiration and tools, the difference between passion and reason, and so on. These laws and laws, AI can not be deeply involved, we must make real efforts! .
For example, science and technology have the characteristics of collective aggregation and continuous relay, and the wisdom of different people can be superimposed and combined. Literature and art have the characteristics of individual dispersion, power discontinuity, etc., the wisdom of different people cannot be superimposed and combined, so the wisdom of outstanding writers and artists is difficult to pass on to future generations and students, talent, understanding, skills and instincts, can not be passed on like genes, everyone starts from the beginning, although it is wasteful, there is no way. Parents are calligraphers, and children do not necessarily know calligraphy. Parents are dancers, and children may not like it at all. The wisdom of literature and art cannot be inherited, but can only be repeated repeatedly, so it is difficult to achieve new breakthroughs at the individual and overall level, and there is even a tendency to regress. For example, writing rhythm poems, no one dares to say that they surpass Li Bai, Du Fu, etc.; Writing calligraphy, no one dares to say that it surpasses Wang Xizhi, Yan Zhenqing, etc.; No one dares to say that he surpasses Cao Xueqin and Luo Guanzhong; Not to mention foreign ones, the reason is that the individual skills of literature and art are not superimposed and relayed. Future generations cannot surpass their predecessors, and AI is even more difficult to do. For individuals, AI can reduce human physical strength and time, reflecting the "principle of minimum action", which is the "principle of God's design of the universe", the multiplication of energy and time tends to zero, the lowest energy consumption, the least time-consuming. But the cost of developing AI is not low.
AI is a fusion of algorithms, computing power, data, and rules, and is a tool for human creation rather than a subject. Similar to today's vehicles and combat weapons, they can far exceed the intellectual and physical strength of individuals, but without emotions, passion, understanding, inspiration and innovation, we do not need fear and worry! Humans are always masters.
Literature and art, in the end, are a matter of people, not things.
(This article is a summary of Li Houqiang's speech at the Chuanpai Review Construction Monthly Symposium on May 28, 2023 at the exchange meeting "The Impact of AI Art and the New Possibility of Art" Forum)