top of page

My Thoughts on AI as an Artist

As AI has become more prevalent in the communities I'm in, I've seen a number of perspectives that lead me to believe that part of our community does not understand the impact this technology has on us. More specifically, I feel the impact AI technology has on artists and other creative members in our communities is often downplayed. Before I can get into why I feel this way, I'd like to share my own stance on AI as an artist and how technology in general has changed aspects of my own work.


When I first started recording and mixing audio on my computer, I found myself refusing to use MIDI. In my eyes, it was inferior to live sound and lacked a certain 'human' element that machines could never emulate. Eventually, I decided to give it a chance and soon realized that a lot of my presumptions on MIDI's capabilities were wrong. With the right VST plugins, I was able to record an authentic piano sound and even tighten up my performance in post. MIDI is simply a tool designed to give musicians flexibility with their work, much in the same way vector graphics did for digital artists and raw camera files did for photographers. So what makes AI technology different from these other technological advances that are objectively beneficial for artists?


Unlike the previous technologies mentioned, there's a growing number of AI tools that are not designed to be used by artists. While tools such as AI Denoising in Adobe Lightroom exist to give more flexibility to artists, the biggest draw to AI technology is content generation. These content generators are aimed at those who want to bypass artists entirely to mass produce content. Not only do these tools harm artists by threatening to replace them, but they also operate at their direct expense.


In order for an AI art generator to work, it needs to be trained on a data set or library of media. This becomes extremely problematic as an artist's works, including those that may have been personal or commissioned for clients, are used to in these libraries without their consent. The AI art generator then produces content that is reminiscent of the artists whose works it's been trained on, enabling others to bypass the need to commission an artist all together. Despite this being common knowledge, for many people, the prospect of owning a personal art machine outweighs the ethics doing so for two reasons; demand and convenience.


In the past, the phrase 'art machine' was used to reference the demanding nature of being a freelance or commercial artist. As an artist's works became more known, the demand and expectations for the artist's services would often grow to overwhelming degrees. Artists described this stress as if they were expected to be a machine entirely dedicated to creating art. To manage this demand, an artist would have to raise their prices and limit their availability either in how much work they would take on or the type of work they'd do. With AI technology today however, that art machine exists and unlike artists, it can keep pace with increasingly high demands. This means there's no waiting for an artist's queue to open, no discussion on what an artist will or won't create, no long waits for work to be completed, and no increased commission costs. It really begs the question, why wouldn't someone chose to use an AI art machine over commissioning an artist?


The most common response I wish to address is that the quality of AI generated content isn't as good as artist-made media. Ultimately, this answer is incredibly short-sighted as it fails to take into consideration that AI art generators are improving by leaps and bounds. Not only that, it fails to take into consideration that there are artists of all varying skill levels in both professional and amateur/enthusiast roles. The truth that many may not want to face is that AI artwork is already good enough and it's only getting better.


Recently, an image was shared on my Discord server that was quite appealing. The linework was well done, color palette pleasing, and the piece did a great job portraying it's subject. When attempting to find the source of the image though, it became clear that there was no single source; the image was AI generated. Under closer inspection, a number of out-of-place details became noticeable, one member even pointing out that it seemed obvious in hindsight. The truth is, it wasn't. These details only became obvious to a scrutinizing eye and even then, not as obvious as one would think. Regardless, the situation brought to light the importance of not only locating the source of an art piece, but of the source itself in the context of art.


The disappointment I felt when I realized there was no source for the art stemmed not from whether I could tell an AI generated piece from an actual artist's, but the fact there was nothing behind it. There was no artist to follow and support, no gallery of work to showcase the growth and stylistic changes of an artist, no artist to admire or be inspired by, no one to communicate or connect to, no purpose, no passion, nothing. To remove one's self from art, an act of self-expression, is to express nothing. My disappointment has nothing to do with the quality of the art itself, but the quality of artistic expression.


When an artist creates, they present a unique perspective forged from the entirety of their life experiences. It's a view of life not through a vague generalization, but the obsessive fascination of an individual observer. This is what makes art so special. It makes us question things about ourselves and the worlds we perceive, it moves us because it says something. When an AI program generates, it produces generalized content that is aseptic, devoid of narrative, ethics, and intent. At it's core, it is nothing but a prompt.


As time goes on and the amount of AI generated content increases, the voices of artists will be drowned in a sea of meaningless noise and self-expression will be lost to general-expression. This is the cost of the convenience AI offers should we accept this into our communities.


146 views

Comments


bottom of page