Artificial Intelligence Invades Art Circle Robot Rembrandt Comes Out

Artificial Intelligence Invades Art Circle Robot Rembrandt Comes Out

AsiaIndustrial NetNews: Maybe many people still think that artificial intelligence is still far from the art world. In fact, AI has quietly entered the chain of curation and creation in the art industry. Once there is a breakthrough in cost, artificial intelligence will turn the art world upside down.


Those areas that have traditionally been considered human-only and AI-impenetrable, such as emotion and creativity, are not insurmountable either. Last year, Microsoft conducted “The Next Rembrandt” project, which challenged the question of “If Rembrandt were to return from the dead, what would he most likely draw”, exploring the creative potential of artificial intelligence, teaching computers like painting Think, create and paint like Rembrandt.

By analyzing Rembrandt’s existing works, the research team allowed the machine to learn the artist’s painting style and theme autonomously. “Robot Rembrandt” finally created a painting.3D printingThe work depicts a portrait of a man in his 30s or 40s, wearing a hat, with a beard, and facing the right. Visitors feel that the Rembrandt hanging in the museum is not at all inconsistent with his works.

Is this painting a creation or a Robot imitation? People in the art circle believe that the “next Rembrandt” is just artificial intelligence based on Rembrandt’s paintings, “imitation and processing” of human creativity.

Since 2014, Tate Britain has established the IK Prize, a science and technology award, to fund projects that use technological means to creatively explore the Tate collection. Last year, the artificial intelligence curatorial project Recognition won the 2016 IK Prize.

Recognition matches the real-time news images provided by Reuters with the collections of the Tate Museum in four dimensions: object recognition, face recognition, composition analysis and context analysis, and then comprehensively calculates the weights of each dimension, and finally selects Generate the most matching news images and collections.

Art from the media “Uncle Tianxia” believes that the current AI curation is still at the stage of sorting out data and data, but in the future, if robots can be trained to be closer to the artistic literacy and way of thinking of human curators, more cultures can be comprehensively analyzed. History, artist background, creative language, and even dialogue with artists, then artificial intelligence will make the art curatorial circle impress.

Can robots be creative too?

when artificial intelligenceindustryWhen people in industries such as agriculture, agriculture, finance, and intermediary love and hate, some people in the art circle disagree, because they think that artistic creation is a brain behavior that humans have not fully understood, and the unpredictable talents of artists. , is far from being replaced by machines.

But in June this year, the McCann Group conducted an interesting experiment in Japan that may make people in the art world reflect. The experiment also had AI robot and human creative director Mitsuru Kuramoto devise a TVC ad creative for a chewing gum brand. Both entries were placed online at the same time, and viewers were invited to vote online for the favorite. In the end, the bot got the same number of votes as the human creative director, losing by a narrow margin of 46% to 54%.

“On this day, robots can write novels, can prioritize their own happiness, and no longer work for humans.” This sentence comes from a science fiction novel “The Day Computers Write Novels”, whose author is an artificial intelligence (AI) )robot. Last year, four works represented by this novel participated in the Japanese “Xingxin – Miniature Novel Literature Award” competition, and at least one of them passed the preliminary review of the competition. Although it did not enter the final round of the competition and won the award, this is the first time that the award has encountered a non-human author who came to the competition since it was established in 1981.

The four artificial intelligence works come from two teams of scientists. The writing process is roughly as follows: Humans first figure out the basic structure of the novel, and input artificial elements such as character settings and content outlines to the computer. Then the computer organizes language based on these “materials” and writes novels.

The process is similar to the AI ​​bots that previously wrote press releases and speeches. Once, WordSmith, a robot created by natural language generation company Automated Insights, competed with American Public Radio reporter Scott Horsley to write a press release on an earnings report. The end result is that robots are more than twice as fast as humans. Comparing the two manuscripts, however, it is clear that human language is more vivid, detailed and emotional.

How do robots get creative? The answer given by Google is that the basic idea is to let the machine learn the rules, recognize and imitate, and then teach the machine to break these rules and make innovations. “Deep Dream” launched by Google in 2015 is a low-profile version of this idea. On the basis of the artificial neural network algorithm, the images input by humans are converted into images recognized by the machine, and the generated images show more animals such as dogs and eyes, and the images will be covered with swirling, rainbow-like colors. Due to the strange style of painting, it also quickly spread on social media.

In 2016, Google established the “Magenta” team with the aim of allowing machines to imitate human imagination and create music and art. The artificial intelligence they hope to create should not only be able to create “replicas” of similar samples, it should have its own style, and only then can it be said to have the ability to learn. The project initially required video and audio-based support, as well as a platform that could connect artists and machine learning models. As you can imagine, with the help of Magenta, the future of art creation will be very simple.

Of course, the difficulty of developing creative robots must also be acknowledged by Dalian Google, which may eventually require human participation. Artificial intelligence for artistic creation, even if it is theoretically possible, needs to be debugged to mature applications, and it needs a lot of specialized training for machine learning art information. For a small-scale industry like art, such a large R&D investment, this It is also unacceptable to commercial organizations. However, in the future, when artists conceive new works, it is completely expected that the robot will do some basic work such as consulting historical materials, and discuss collision inspiration with the robot.

Looking forward to the future

It is found that artificial intelligence is slowly evolving from “replacement of blue collar” to “replacement of white collar”, such as corporate executives, tax advisors, stock analysis, and even college teaching assistants. Jerry Kaplan, a professor of artificial intelligence and ethics at Stanford University and author of “The Age of Artificial Intelligence,” believes that some low-skilled, monotonous, repetitive jobs will soon be replaced by artificial intelligence. However, some work environments are complex, and solutions need to be proposed in real time and quickly adapt to social changes. Such work is safe.

Kai-Fu Lee, founder and CEO of Innovation Works, believes that instead of considering whether artificial intelligence will enslave human beings in the future, it is better to first consider whether artificial intelligence will lead to the unemployment of a large number of people in the next 10-15 years. Although the development of artificial intelligence will also promote the creation of new jobs, the jobs created are definitely far less than the jobs lost, so it is a big challenge for human beings.

However, whether robots can be competent for jobs that require “sensibility” and “creativity” in the future cannot be concluded. In the field of literature, compared to press releases describing information and facts, novels with more literary skills have higher requirements on words and thinking. After all, a novel must not only be logically rigorous, but also emotionally rich and well-characterized. Excellent novelists don’t have to worry about their jobs for the time being.

In the field of art, artificial intelligence can theoretically do more rigorously and comprehensively than humans. For example, image analysis, art history research, auction database, art market analysis, artwork identification and judgment, etc. Machines learn images or texts autonomously, then classify and summarize, and analyze and predict, which can reduce many unnecessary or inaccurate judgments. However, “Uncle in the World” thinks that this sounds very good, but the cost of hindering the application of technology is very important.

Gu Xianfeng, a tenured professor at the State University of New York, believes that the method of artificial intelligence can separate artistic content and artistic style, and can accurately represent the content and style, and freely convert and synthesize the artistic style. Although the capabilities of computers are astounding, the aesthetic and judgment of works of art today is still irreplaceably performed by humans. So, will in the near future, human aesthetic ability will also be quantified, accurately deduced by mathematical formulas, and eventually replaced by artificial intelligence?

The Links:   3HNA023093-001 3HAC7343-1

Published on 08/30/2022