Fanessay整理了一篇paper范文--About artificial intelligence painting，本篇文章讲述的内容是关于人工智能绘画方面的，可以供大家了解下。在2018年，人工智能开始利用算法程序挑战艺术创作。2018年10月，Obvious法国艺术团体AI的绘画作品《埃德蒙·贝拉米像》便在佳士得艺术品拍卖会上压轴出场了，并且拍出了天价。同一年，谷歌发布了自主学习绘画的AI程序“DeepMind”，人工智能开始学习大量图像，进行分解、重组与再创作。这也让很多对于人工智能绘画的能否可以有创新价值开始了新一轮的争议。
Artificial intelligence began to infiltrate into different fields, and gradually specialized in each field, which was regarded as an "artistic creation" that could not be replaced, and was inevitably impacted by artificial intelligence.
In 2015, artificial intelligence began to use algorithms to challenge artistic creation. At the beginning of the research, these artificial intelligence was only able to copy masterpieces of famous artists rather than make original paintings. And 2018 can be regarded as the year when artificial intelligence breaks down barriers and formally enters the art world. In October, AI painting "the portrait of Edmund Bellamy" by Obvious French art group came to the end of Christie's art auction and sold for a record price. That same year, Google released the autonomous learning painting "DeepMind" AI program, artificial intelligence, begin to learn a large number of images, decomposition, restructuring and creation, which makes many scholars for artificial intelligence can painting creative value with a new round of debate, people began to explore the creative process of painting of artificial intelligence, trying to find the similarities and differences between AI and paintings and human.
The development of artificial intelligence needs three elements: big data, software and hardware. Artificial intelligence learns through big data. Hardware upgrades will improve the computing power of the computer, and software will become stronger. It can be said that the core software of artificial intelligence is deep learning, which allows artificial intelligence from concept to implementation and simulates the learning mode of human brain. Through massive data learning, tens of thousands or even hundreds of millions of steps are calculated every second. When Alphago beat the world's top go players, many people were amazed at what the deep learning capabilities of ai could actually achieve. So, ai painting and ai chess thinking mode is consistent?
Here to "Edmund Craig Bellamy like" as an example, it USES the generator in GAN to learn the rules of painting, all kinds of properties characteristics of objects, in accordance with these rules attributes to create paintings, reusing created another discriminator analysis of images from gallery, or a machine generated, once through the discriminator, can output image, moreover the team add randomization module to ensure that the uniqueness of paintings of artificial intelligence. Extending this rule to Google's "DeepMind" analysis, its drawing algorithm is similar to Obvious GAN. Artificial intelligence, the generator and discriminator, is equivalent to playing two roles simultaneously: student and teacher, creator and connoisseur, which is a stage achievement for AI painting.
There is no denying the fact that there are some scholars don't think the AI painting is an art, its work more cannot be called art, because art is considered to be the creator of the understanding of the objective world, is the subjective emotion, the art activity is a creative process, it is full of emotional colors, the greatest feature of the human art creation is the emotional; However, artificial intelligence is rational, and its whole art production logic is based on big data. Even if randomization module is added, the painting content is still extracted, decomposed and combined from a large number of works, which cannot be called sentimentalized art creation. Foreign scholars also believe that there is no possibility for artificial intelligence to create works equal to or greater than human intelligence, and such highly personal creative activities cannot be copied.
Some scholars believe that artificial intelligence takes algorithms as art language, information processing as art means, deep learning as art creation process, and finally forms meaning in symbolic form. This whole set of new artistic creation forms seems to have the creation conditions required by artists, and also conforms to the process of artists' learning. Intelligent machine deep learning is used to create various kinds of artists' creation, through "0" and "1", to create similar or cross-style paintings. In short, these artificial intelligence machines are based on powerful databases, use their own editing programs to process information, and make final input after meeting their own evaluation criteria. This new form of artistic creation has brought impact to many artists.
There are quite a few similarities between artificial intelligence painting and artist painting. The similarities here include learning theory in the early stage, copying by famous artists in the middle stage, and even creation with personality in the later stage. In the early stage of the theoretical study, has a large capacity storage intelligent machines are superior to human learning speed and breadth of knowledge, there is no relief phase intelligent machines that can grasp the characteristics of the famous painting in a short time and fast, high quality copy, you can imagine, part of the restricted by traditional techniques of art is replaced by intelligent machine beyond or easily.
DeepMind has created a series of absurd and irrational works, from which we can conclude that the world as seen by ai is not yet regulated and that intelligent machines have no fear, not only in their actions but also in their "thinking" output. However, in the process of human art, the paintings of modernism and postmodernism are also absurd and irrational. They are artists' re-interpretation of space and time in that period, as well as a new way of thinking and expression. Therefore, we cannot deny that artificial intelligence has another way of interpreting the world.
In 2017, Google launched "AutoDraw", in which users complete their doodles by several strokes, identify the contours through the smart program, and convert them into the images required by users. The images here are the corresponding images generated based on the features of different doodles, which breaks the rigid definition of things by users and facilitates users to innovate in visual images. A machine or an application on a mobile platform makes art more accessible to people, and artificial intelligence makes the idea that anyone can be an artist more realistic.
Need not pessimistic, we must realize that the most impact is embodied in the physical plane, for now, artificial intelligence, it is hard to replace the artist's painting creation, even if these intelligent machines to create some recognized works of art, it is based on the people involved in design, this kind of unconscious creation cannot be separated from human. People need to understand the painter's personal life, social, political and economic conditions, in order to try to guess the meaning of a picture, and artificial intelligence painting is still difficult to make the work full of such emotional color.
So, what's the point of the art of the present technology? A painter's decades of skill practice are as effective as a machine's short period of data learning. Does that mean that decades of human learning are useless? Actually otherwise, anything must depend on the extensive work foundation, this human painters and intelligent machines, but the painter painting creation is a personal aesthetic based on learning experience and through the combination of image rendering, painting can be the audience recognition, must be to rise to emotional touch, for the machine, right now no one can give it desires, we have to at this time of the paintings of artificial intelligence as an experimental or a commodity.
When discussing whether artificial intelligence can replace the creation of artists, we might as well take artificial intelligence painting as the key to inspire artists. Admittedly, ai painting gives artists a sense of urgency, but this pressure is not necessarily a driving force. In reality, the creation of an artist will be affected by a variety of factors. The solidification of thinking and the lack of long-term inspiration are the biggest stumbling blocks for an artist on the road of innovation, which no artist can avoid. Artificial intelligence painting works can be crazy, can be absurd, the same work can also expand the artist's thinking blind spot. The restructuring of the world famous painting "Mona Lisa" in artificial intelligence, with a variety of presentation, it can combine with "star" star version of the Mona Lisa, it can make use of motion sensors "resurrection" "be pull lisa", these bold attempt and flexibility, more or less can bring artists are either positive or negative thinking.
Designer qu minmin believes that the definition of any thing is uncertain at present, for example, whether a designer is an artist should be determined by later generations. The famous painter van gogh only painted after his death. Similarly, whether the painting with artificial intelligence is an artistic creation may need to be judged by later generations. At present, artificial intelligence has reached a degree of specialization in many fields, and people even start to rely on the judgment and instructions made by intelligent machines. More and more artists are also beginning to learn to make technology cooperate with their works. At present, perhaps we can let the painting of artificial intelligence as our second inspiration, so that technology and art burst out of infinite possibilities.
The era of artificial intelligence has come, and its stage achievements in the field of painting are not enough to frighten artists. At present, the artistic creation of artificial intelligence broadens the horizon of artists and stimulates their potential. It is not difficult to find that more and more scholars are interested in the research on "how to integrate artificial intelligence with art", which not only proves the universality of artificial intelligence, but also reflects the infinite possibility of future art development. Science and art are not separate, what we should do is to find a better channel, let art in the way of science and technology, let art shine. After all, you won't be replaced by an intelligent machine, but by someone who can harness artificial intelligence.