Engineers are pushing the boundaries of creativity, creating a new medium and raising questions about what art can beSource: © Can artificial intelligence create art, or is it just an ersatz?
At London's Tate Modern museum last month, visitors roamed the exhibition halls, appreciating the multicolored brushstrokes with which French artist Paul Cézanne created his still lifes, portraits and landscapes, and s marveling at his creativity, talent and impact on XNUMXth century art.
But today, more than a century after his death, artificial intelligence (AI) technology is able to reproduce his work in just seconds.
Indeed, today's AI systems can create any image in any style - from Impressionism to Cubism to pop art. All they need is detailed code.
The world of creation is full of potential. Previous advancements in AI have allowed computers to compete with humans in many analytical areas, leaving the creative work to artists, writers, and designers. Now, however, the new field of “generative AI” gives machines the ability to create entirely new works, drawing inspiration from the immense amount of online data and knowledge accumulated over centuries.
According to experts, this phenomenon could revolutionize human creativity, forcing professionals, whether software engineers, writers or artists, to radically rethink the way they work.
The incursion of AI into the arts raises questions about the importance of human input in the creative process. Can art exist without an artist? If a work is created by a machine, who owns it? What are the hidden dangers for society and humanity? the Times of Israel spoke to experts to try to shed some light on some of these questions.
The overall conclusion is that the creative industry has no choice but to embrace AI. Rather than crowding out human artists, new technology will work with them to create new types of work, pushing their inventiveness and creativity even further and creating something entirely new together.
“Painting by codes” in the age of AI
Artificial intelligence – the technology that gives computers the ability to learn – has been around since the 1950s. But over the past decade the field has seen a renaissance, made possible by the huge amount of data available online. and the increased computing power of the chips. Advances in this field over the past decade have allowed computers to analyze all accumulated data and find useful patterns to solve certain problems, with the machine often outperforming the human brain. Creativity, however, remained primarily in the hands of artists.
Today, software such as ChatGPT, GIVE HER, Midjourney and Stable Diffusion advance creativity by leaps and bounds. And while these tools were previously only accessible to researchers and a small group of testers by invitation, a revolution is now underway. They are made public and can be easily used by everyone, in what some call a “democratization” of creativity.
On November 30, OpenAI, an AI research company co-founded by Elon Musk in late 2015 and backed by Microsoft, widely released its ChatGPT bot prototype, capable of generating sophisticated texts in response to prompts and questions. The robot has attracted more than a million users in five days, according to the president and co-founder of OpenAI, Greg Brockman, and has won over its users with its potential to disrupt research work, journalism, code writing, academic papers, literature, and more.
Other software is also making headlines in the world of art and design: DALL-E, developed by OpenAI and made available to the public as part of a beta trial in November; Midjourney, which released a beta version available since July; and Stable Diffusion, which developed a text-to-image deep learning model, also released this year.
These new AI-based image technologies can create people, objects, and places and mimic entire visual styles based on user requests or prompts. By following simple written instructions, the latest version of DALL-E can create, for example, realistic images of a polar bear playing bass or a Picasso-style robot.
The fact that these services, once reserved for a closed circle of researchers, are now accessible to the general public is already having "a significant impact on creativity and culture", said Asaf Hanuka, director of the department of visual communication at Shenkar College. of Engineering, Design and Art by Ramat Gan. "All of this is happening right now," he said in a phone interview.
Shenkar, who held a talk earlier this month in Tel Aviv on Creativity in the Age of AI, highlighting the pros and cons of the technology, is already seeing students using the tools in their graduation projects. year, Hanuka said.
Art or artifice?
Artists have been experimenting with AI for their works almost from the advent of this new technology.
Artist Harold Cohen, a pioneer of computer art, developed an art-making program beginning in the 1970s called AARON – "the world's first pure AI artist", according to the Description of one of his works at the Victoria & Albert Museum. Cohen's software, however, was not a open source and was therefore not accessible to the general public. Additionally, AARON's development ended with Cohen's death in 2016.
In recent years, however, AI in art “has exploded,” said Drew Hemment, professor of Data Arts & Society at the University of Edinburgh and fellow at the Alan Turing Institute. This is due to both "advances in technology, and the release of powerful new tools".
“AI gives artists superpowers,” Hemment said in an interview via e-mail. “Today, artists can create images, or sounds, or anything else they can imagine that harnesses the creative powers of artists who came before them, and merges human intuition with advanced computer technology. »
"Today we see artists working with artificial intelligence in very creative ways, both as a tool and as a subject," he added. “Artists are at the forefront of pushing the boundaries of this technology. »
In June of this year, Cosmopolitan used DALL-E 2 for generate the world's first "artificially intelligent magazine cover", with a prompt that instructed the software to create a "lower-body wide shot of an athletic female astronaut walking confidently towards the camera on Mars in an infinite universe , an artificial digital art”.
The results were used for the cover of the magazine's AI issue.
In the United States, Jason Allen won the first prize, ahead of 20 other artists, for his " Space Opera Theater at the Colorado State Fair Fine Arts Competition in late August of this year. Much of the artwork had been created using the AI tool Midjourney, but the judges couldn't tell. This sparked a huge debate over the meaning of the art, and Jason Allen was accused of deception.
"We're in the midst of a very, very rapid evolution, maybe even a revolution, of machines seeping into these creative artistic fields," said Yoed Kenett, assistant professor at the university of industrial engineering. and management of the Technion-Israel Institute of Technology. “Does that mean it's the end of the artistic profession? I do not think so. It just changes what it means to be an artist. And I think that's a great thing. »
“Creativity is the ability to connect, to create new combinations, to generate new concepts, from old ideas, to make analogies, metaphors,” Kenett said. “All of this combined allows us to create this magic. »
Can computers get an artistic license?
According to Wikipedia, "art is an activity, the product of this activity or the idea that one has of it, which deliberately addresses the senses, the emotions, the intuitions and the intellect".
So, can art made by a computer be considered art? Can a computer become the artist?
“For me, the tools an artist uses makes no difference, I'm only interested in the quality of the art. We can make art with a brush and an easel, or with data and algorithms. It's art if it touches us, if it's aesthetically interesting,” said Drew Hemment of the University of Edinburgh.
"Rather than removing the human from the equation, AI will be a new string to the artist's bow," explained Hanuka of Shenkar.
“In creativity, there are two main questions: why we do something and how we do it,” he said. “The computer does not create art. He cannot choose. The 'why' remains the sole domain of the human artist, 'the machine serving as a very eager helper and assistant'.
Art is always "a dialogue between the creator and his working tool", he added.
Israeli artist Oren Eliav, who lives and works in Tel Aviv, has exhibited his works in galleries in Israel, France and Italy, the United States and the United Kingdom, among others. His first solo exhibition at the Israel Museum was inspired by “The Death of Lucretius” by the 20th century Italian painter Giovanni di Paolo. In his work, Eliav deconstructed the original painting and created XNUMX large images focusing on different sections of Di Paolo's composition.
"As an artist, I always work from a DNA that is not mine," he said in a telephone interview. “I always ask a question from an image that I didn't create myself. My artistic work is to question, it is a dialogue with the work of the other artist. »
With the coronavirus restrictions in 2020, Eliav began learning about the AI imaging tools available at the time. Using Generative Adversarial Networks (GAN) technology – one of the first attempts at image generation, which emerged in the mid-2010s – he created a sequence of evolving forms, which served benchmark for his hand-painted works in oil on canvas. The paintings, along with an original poem, resulted in the exhibition “The Moon is a Mirror” at the Braverman Gallery in Tel Aviv, from January to March 2022. The project continues as a “work -video” (long-form), which will be released next year.
“In my work, what I ask of the computer is to start imagining. Then I match the computer's imagination to mine, and something that I can't anticipate in advance is created, which puts me at a very interesting point as a creator,” he said. -he declares.
The AI technologies available today, he said, are an additional tool in his creative briefcase.
“When a person is creative, and they create, they use all the tools available,” he said. “If a person is not creative and not original, even if he has the most extraordinary tools, he will not be able to achieve an original result. These technologies are like a musical instrument: it all depends on the person using them. »
Hidden Dangers
Despite advances, these new AI technologies come with pitfalls. Subjectivity is one of them. All AI systems depend on the information they are fed – and so they could perpetuate any existing biases and biases from which they are formed. Facial recognition software, for example, has been criticized by civil liberties advocates for being biased against people of color, due to skin type and gender biases built into the systems.
To prevent abuse or the perpetuation of violence, limits have been integrated into the image generation tools via the AI. For example, OpenAI's ChatGPT uses a moderation API (application programming interface) to warn against or block certain types of dangerous content. The company has also limited DALL-E 2's ability to generate violent, hateful, or pornographic imagery, as well as political content, according to its website.
“But that will limit the level of creativity in the software,” Hanuka said, because the art can actually depict violence and death and is often a political statement. “'Guernica' by Picasso is an image of war,” he remarked.
Furthermore, according to Hanuka, since these image-generating tools via AI collect and synthesize information from existing content, their natural tendency will be to work with images and elements that are familiar and common – which can lead to predictable and boring results.
"Algorithms prefer the familiar, and the nuances tend to disappear," Hanuka said. “And in reality, innovation often stems from nuances. »
There is also the issue of intellectual property. Art via the AI creates images using models that have been trained on the original work of other people and other artists. This raises ethical and practical questions.
“One of the main concerns is that the AI models currently used by industry have been trained on massive data sets taken from the internet, such as images or information. This is done without permission, without awareness and without fair compensation to the original creators. This has yet to be investigated by the courts, but it is clearly a mistake,” Hemment said.
“I expect to see a new generation of models trained on licensed footage, but this poses a new problem because only a small number of big players have the scale to do so. This leads to increased centralization, with an ever-shrinking group of companies calling the shots. »
The internet and social networks were hailed early on for their ability to democratize information and make the world more connected. Yet both have also enabled invasion of privacy and political manipulation.
“The lessons learned over the past 20 years must now be used to ensure the legal frameworks are in place to tackle image generation tools via AI,” said Tehilla Shwartz Altshuler, policy officer at the Israel Democracy Institute (IDI) and expert in law and new technologies.
"We know today that these different platforms, which allow us to create our own content, must have some form of responsibility for the result created on top of them," she said. “That's happening today with social media platforms, but the next frontier will be image-generating tool platforms. via the AI. These platforms will not grow with impunity, because it is 2023, not 2000. We have learned from our mistakes over the past twenty years. »
“Restrictions on pornography, terror and violence should be enforced on these platforms,” she insisted, “and detection systems should be created to help us tell truth from lies and show us where, when and from what source content was created”.
The surprise effect
To test the technology, a journalist used Dall-E to create two images.
“After registering with a username and password, I entered the following prompts: female journalist sitting at her desk with a laptop and a hot cup of coffee in the cubist style. »
Within seconds, four images were created.
“The second message, to please my Dutch husband, was: the Dutch football team rejoicing in the World Cup victory in the style of Van Gogh. I plan to give him this image – created in seconds – as a consolation prize for his team's loss. »
So, whatever the hidden dangers that these new technologies may harbor, it is clear that AI is well on its way to persisting in the creative sphere. Just as we've largely ceded our browsing capabilities to Waze and our information gathering to Google, we're going to cede more and more of the creative process to the machine.
"Technology will enable more people to be creative," said Technion's Kenett, "by giving them tools to express themselves, even if they can't write or paint well, which will allow a democratization of creativity”.
And the question of whether or not AI has been used, to create a work of art, will quickly become superfluous, Shenkar's Hanuka said.
"Whether or not you used a computer is irrelevant and that question will go away over time," Hanuka said. “It doesn't really matter. What matters are the results achieved. A creation that pleases, that creates an impact, that's what we will remember. »
