Jason Allen, the creator of the painting “Space Opera House”, is not a professional artist, but a game designer in Colorado, USA. This work was made by him with artificial intelligence (AI). Previously, whether it was “Deep Blue” playing against the Russian chess king Garry Kasparov, or “AlphaGo” defeating the Korean Go player Lee Sedol, human beings can still regard AI as a computer with powerful computing power, rather than a machine with human nature, because human beings I always think that philosophy and art are unique fields that are different from AI, but when AI has begun to enter the field of art, people can’t help but ask: Will AI replace humans?
destruction or progress
In human history, industrial technology has always been inextricably linked with art, but the technology at that time either promoted the development of tools such as brushes and paints, or promoted social progress, influenced people’s world outlook and values, and thus promoted Art produces more genres, but there has never been a technology that directly produces works of art. Therefore, some people exclaimed: Human art will cease to exist, and AI may replace humans.
Whether AI can replace humans, Zhai Zhenming, a professor of philosophy at Sun Yat-sen University and director of the Human-Computer Interconnection Laboratory, believes: “Art will become more pure in the future, and AI can help us filter out real artists.” In Zhai Zhenming’s view , the innovation of this technical element will push the originality of art to a higher position. “AI can replace craftsmen, not artists. The repetitive and delicate product generation work is handed over to AI, and artists no longer need to pay for it. What they need to do is purer artistic creation.” This is because art It can be divided into popular art, commercial art and avant-garde art. AI creation can replace popular art and has sufficient potential in commercial art, but it is difficult to replace avant-garde art.
Popular art is aimed at the aesthetic needs of the public, such as New Year pictures, portraits, animation images, etc. AI creations are outstanding in this field, and many painters make a living by doing so. Leng Jun is a hyper-realist painter , whose masterpiece “Little Ginger” depicts an image of a woman with exquisite details, which is called “a painting made with a camera”. Using a text-to-image generation model, AI can also create a satisfying portrait. “AI portraits completely capture the essence, big and small eyes, no nose bridge, wide eye distance, square face, pointed chin and high cheekbones,” a Weibo netizen commented on the AI’s portrayal of himself. AI also has sufficient potential in the field of commercial art. Commercial content such as logo design, clothing design, and packaging design can be completely designed by AI to meet the business needs of enterprises. Liu Chang, a self-media person in Datong, Shanxi, once admitted in an interview that most of the pictures on their WeChat public account were found on the Internet or customized by an artist. They spent a lot of time and money but rarely met their requirements. After buying the free AI painting tool, I finally don’t have to worry about it anymore.
But in the field of avant-garde art, AI does not have human feelings and experience, and cannot deeply interpret the content and meaning of works, neither can it open up new forms of expression, nor can it create new genres out of thin air. In 2017, CCTV launched a program “Extraordinary Wit”. In the program, AI Xiaobing and human poets competed on the same stage. Facing a photo of Jiang Yiyan going to the countryside to support education, the two poets and Xiaobing gave their opinions. work. In the works of human poets, human feelings and stories are clearly highlighted, but although the works of AI Xiaobing are full of images and metaphors, they don’t understand the protagonist Jiang Yiyan’s teaching experience, nor do they jump out of the pictures to think about the future, so the works also appear Distant and icy. It can be seen that AI may be able to write a good poem, but it is difficult to write a great poem.
plagiarism or creation
As AI painting became popular on the Internet, some painters said on Weibo that many of the pictures generated on an AI painting platform are completed in the form of “puzzles” of his works. This is because the current AI painting still adopts the method of machine learning in the final analysis. With the support of the algorithm, by learning a large number of existing paintings, a certain ability of recognition, classification and combination is generated. When people input some keywords, It will find the characteristics of related paintings based on these classifications, and then generate a work. Therefore, various familiar elements can be seen in many AI paintings, such as anime characters, architectural scenes, game backgrounds, etc.
However, whether AI painting constitutes plagiarism still needs to be analyzed in specific cases. Cao Ying, a lawyer from Beijing Global Law Firm, said in an interview: “As a basic principle of copyright law, the ‘ideology and expression dichotomy’ has guiding significance in judging whether an AI painting is infringing. If the elements of the original painting are too abstract, High, it may cause this element to be identified as an ‘idea’ not protected by the copyright law rather than an original ‘expression’, then the AI painting does not constitute infringement. In other words, the ‘shattering’ AI painting is more similar to an original expression, It is difficult to constitute infringement; the degree of processing of ‘Meitu-type’ AI paintings is relatively shallow, and its possibility of infringement is relatively high.” Huang
Qunhui, a lawyer from Beijing Yinghe Law Firm, said in an interview that learning other people’s works for non-commercial purposes is legal. It is allowed in the above, that is, simply using existing picture works for learning itself does not violate the law. As a copyright owner, a painter has the right to prohibit others from using his works for profit, but he cannot prohibit personal learning, research or appreciation, which is also the legal restriction on the author’s copyright. Therefore, he believes that there will be no copyright disputes between AI paintings and painters’ paintings in the future, and the two will develop in different directions: In the future, intelligent AI painting software will no longer copy and paste, but its own unique In some painting databases, these data can be arranged and combined to form different works for screening. The works of painters are on the contrary. In addition to creative skills, they present more humanized expressions of thinking to the audience, and the focus of their creations will be completely different. Therefore, AI painting will be an independent art form.
Some people think that human learning and AI learning have a similar process, but some of the mechanisms cannot be explained by existing science, so there is a difference between humans and AI. Taking a painter as an example, human learning also needs to learn the difference and use of colors, appreciating works of various artistic styles and paintings of famous painters, and finally integrate these into their own materials and form their own style and characteristics in painting . However, people have not yet figured out the mechanism of human beings in information processing and material processing, so AI and humans show a huge difference. This difference is often called human emotion. Therefore, with the development of life sciences, AI learning may be comparable to humans.
Algorithms Win and Humans Hold Back
When “AlphaGo” began to learn by itself and generated game records, people were surprised to find that the masters that humans have summed up in Go in the past turned out to be ordinary players in the eyes of AI. Many moves have completely surpassed human understanding. Jiuduan said: “I have never seen these moves before, I think this is from the game records of the future”, while another world champion Gu Li Jiudan said: “AlphaGo’s self-trained game records are unbelievable, we can learn from Learned a lot.”
