Fascinating or jarring? A step forward or a step back? These questions, words, and phrases all describe one of the most hotly debated topics today: A.I. art. Art that is produced by artificial intelligence programs has fostered many perspectives around the world. Some ranging from positive to some negative. What makes it so controversial, and what purpose can it really serve to artists and to other professions alike?
One of the first major instances of A.I. art breaking into the mainstream was with the Portrait of Edmond Belamy in 2018. This portrait was really a part of a collection of portraits that was put up for auction by the French-based company Obvious. The entire collection was generated by A.I. and sold for up to $432,000. This number, about 45 times the estimated price of listing, baffled many – from critics to normal art connoisseurs. Most critics thought the portrait was underwhelming for its perceived insane price tag. This opinion of the underwhelming qualities of A.I. art has persisted until the modern day with even more discussion being brought up about it.
The collection of portraits was generated by a machine-learning algorithm known as the Generative Adversarial Network (GAN). Using a database with selected 14th-20th century portraits, GAN generated a portrait until its discriminator system could not tell the difference between its generated portrait and a real portrait from that era. This system helped create the portrait that sold for almost half a million dollars. The algorithms powering this kind of generation and discrimination, as it’s called, has since evolved. Generally, all one needs to do with today’s systems is to type in a prompt into generators like DALL-E2, Img2Go, or Midjourney, and the generator will create an image on the spot. Typically, these generators offer a selection of images for the user to choose from. The generation is a result of a selection of data from a database, a training of the machine to learn features and patterns that are common in art pieces, and a refinement system that can filter and change styles of the artwork among other things.
However, the database that powers these systems is one of the most controversial aspects of the way A.I. art is generated and made. The databases should theoretically compile public domain images, but copyrighted images often find their way into these generator’s databases. Copyrighted images are taken from artists who never knew, never consented, and/or are never credited for their work. Artists are not being compensated by the generators who take their work, and their work is being used by the A.I. to make its own version. Some generators are so obviously using stolen art that remnants of the original artists’ credentials can be found in the generated work, typically in the bottom-right or bottom-left-hand corners. A possible solution to this conundrum would be to simply not use copyrighted images without the artists’ expressed consent. If an artist does consent, a royalty should be compensated for the artist.
The accessibility of these A.I. generators and their ease of use creates other serious problems: the problem of replacing real human workers who illustrate, draw, or even animate. With the emergence of A.I. art, it is plausible that any one company or employer would use it for their advertisements or their forms of art instead of hiring human workers. This possibility is what sparks the worry of the removal of these jobs. Getting into the art field and finding success is hard enough, but to have that opportunity completely taken away from people who aspire to be and currently are artists is something that does not sit well with them.
On that note, A.I. art does have its own distinct “feel” or style, or perhaps its lack of a well-defined one. Despite the fact that A.I. art generators usually have multiple styles to pick from and have the user type out a custom font, much of the art that is produced feels strangely surreal, or perhaps too real, or incorrect in its proportions, the perspective, details, etc. Art often seems good, but upon closer inspection visual inconsistencies become apparent quickly. The most notable controversy surrounding the A.I. art’s peculiarities were with hands. All over social media and the internet, people were pointing out these visible peculiarities as a way of mocking the A.I. art and finding people who used it on their own posts. The hands of A.I. art became almost a trademark of all the generators as they struggled to comprehend the proportions of the hand that even perplexes human artists.
If the creative style of the art is lost, then so can the creative process. Many artists today often say that A.I. cannot be creative because it lacks the humanity in making art. It does not have to go through the stages of coming up with an idea, planning it through, and making adjustments. It does not include the process of creating the art. Usually, the process ends by the artist achieving something, but the A.I art does not have this feeling of achievement. Many say that this process, along with the actual art, is what the point of creativity in art is. Simply running through a program’s algorithm does not yield any emotional reward.
Although the generated art may or may not be creative, that does not mean that A.I. art cannot somehow help others to find their creativity. Some people mention how they find A.I. art helpful when they need to find an inspiration or a catalyst of an idea when they cannot find something that already exists online. Generating the idea into something like DALL-E2 could provide someone with the ideas they need to further their journey on completing their one particular project.
A.I. also does not necessarily have to impede the creative process, but can condense it. A.I. could help simplify the “busy work” or tedious tasks that artists often have to do in their process of completing a work. Especially for animation, where numerous frames have to be done by a human that may be repetitive for many. The A.I. can be trained in that animator’s art style and can help speed up the process of finishing an entire animation. Some animators, however, enjoy the prospect of continuously drawing the same or similar things because it is part of their creative process that they value immensely.
In this sense, A.I. art could be used as a sort of collaborative tool for artists instead of a tool that may work against them. This tool could help them find inspiration, do repetitive work, as well as act as a sort of “sketch” tool that can get preliminary ideas on paper. This type of visualizing for someone can be applied to jobs like architecture who may need some type of visual of their imagined building, or how a certain part could attach to a building, and so forth. Another job that could potentially benefit is that of designers. For example, interior and exterior designers could have the A.I. sketch an interior or exterior of a home according to the customer’s request, and the customer would have ample ability to edit and regulate what they want and what they are receiving from the generator. This visualization can help professions that regularly require visual mock-ups of their planned future work, like architects, and interior and exterior designers. A.I. art could act as a first step in a long complicated process of finishing a work of art.
This collaboration prospect has already been changing the film and T.V., specifically on VFX and CGI. With A.I., CGI artists can create more realistic visual effects, digitally age someone older or younger, and create characters with the use of an algorithm. A.I. is also behind the infamous deepfake technology that could potentially replace actors with their likeness and image, assuming that they would consent to it. Deepfakes, in particular, are a controversial topic in their own right as they raise the question of validity and truthfulness. A public figure’s image and likeness can be easily manipulated and faked online, which can cause many to believe the falsity. Still, these technologies do have the potential to do many elevating things, but many feel, similar to A.I art outside of these industries, that it reduces the creativity of CGI and the whole process of creating CGI.
A.I. art as whole is not only an artistic dispute, but an ethical one. The potential for it to help artists to succeed is there, but just as likely is its potential hindrance to them as well.