Is AI art truly art?

Art is as essential as it is mysterious. Art – the capacity to both make it and appreciate it – is one of the characteristics that make us human, but it is also very difficult to explain what we mean when we say “art.” We can see this whenever a new medium that purports to be art achieves prominence, and is immediately subjected to that age old debate of whether or not it qualifies as art. This has happened with fashion, with abstract painting, with comics and with video games.

The reason for that, I believe, is easy enough to understand. While art is a concept universal to humanity, it is experienced on a very personal, subjective level. One of the ways that art has been defined is a fundamental tendency to make something “special” or to respond to “specialness.” But what seems special can vary from person to person, from context to context. This means that any attempt to define art immediately runs into the problem of the definer’s personal taste of what is “special,” or what is “good enough” in an aesthetic sense, to call art. To put it another way, Leo Tolstoy once defined art this way: “Art is a human activity consisting in this, that one man consciously, by means of certain external signs, hands on to others feelings he has lived through, and that other people are infected by these feelings, and also experience them.” What resonates with people on an emotional level is different for each person, so coming to a common consensus on exactly what art is would be very difficult.

Historically, the element of the human in the maker of the art was uncontested and unexplored. However, recent advances in technology have seen the rise of so-called generative AI (artificial intelligence) systems which – when given the barest instructions or ‘prompts’ – are able to craft images and text at a level that, for many, seems human made. Some have heralded this as a revolution, a “democratization” of art that allows those without skill to mimic the proficiency of a seasoned writer or artist. Many creators, however, have raised issues regarding how these AI systems were created, and how their unchecked use could harm artists.

The artists raise valid points. While commonly described as “AI Art,” the generative systems that create these works are not “thinking” machines, such as one may imagine from science fiction. They are “machine learning” or “deep learning” systems that use algorithms to create something out of data samples that they were given and “trained” with. Programs that make art were trained with art – the problem, however, lies with how this art used to train the systems was obtained. Many of the most widely used AI systems did not obtain consent, nor provide compensation to, the human artists that made the art their systems were trained on. One of the most common myths about content found online is that if one can access it, it is free to be used however one wants – this is patently untrue, as copyright still protects work on the internet. But the creators of many of these systems were privy to that; they simply did not want to do the right thing because, as the founder of Midjourney has implied, they found it to be too difficult to do. Of course, both legally and morally, that is no excuse.

What we have currently are many generative AI systems that were knowingly trained on the copyrighted works of artists and writers, with neither consent nor compensation, and which systems are now being used to generate profit and content for other parties. This is both a violation of copyright laws and an exploitation of labor – and yes, labor in the fields of the arts is still labor. And under our Constitution, the “State affirms labor as a primary social economic force” and “it shall protect the rights of workers and promote their welfare.”

There is certainly a space for machine learning systems to serve as tools and aids for creation, for those that choose to use them. But these systems cannot be built using means that violate the rights of the artists who create the images and words that enable the systems to learn. This means creating explicit legal protections for our creators, and preventing the monetization or regulating the use of existing systems that were created using exploitative means. I encourage our Congress to consult with our artists and writers – bearing in mind that generative systems impact creators at every level. This is necessary in order to best protect the artists’ and writers’ rights and creations.

Art can and does evolve, and the tools that artists use also change. But in as many forms that art can take, it is fundamentally the inner world of one human connecting with other humans. Generative systems may be able to mimic and iterate what humans originate, but they cannot be “inspired” by them, they have no emotions or feelings to share or express. They only take from what has already been produced by a human, so it can be argued that they cannot truly produce something new – something never been done before. And novelty, that element of surprise, is essential to the evolution of art.

While art can and does exist as a product, we cannot allow it to only be valued as a commodity. Those concerned with art as a product must contend with the proper legalities, with the realm of copyright and compensation. Those who claim to be concerned with art and its so-called “democratization” – as well as those concerned with ethics – must always be on the side of the artists, of the cultural workers.

Images and words may be generated in the millions, but there can be no art without artists. On any scale of value, whether ethical or cultural, the welfare of the artist must reign supreme over the production of art.

There can be no revolution for art that devalues artists. As we celebrate National Arts Month this February, let us keep this in mind: those on the side of art, must be on the side of artists.