Dairn Alexandre (a pseudonym) is a regular BILD guest blogger; for more information about Dairn, and to read his earlier posts, click here. Dairn has taught in Quebec and now works as a teacher in Alberta, where he lives with his wife, two kids, and dog.

The image above was created using artificial intelligence (AI), specifically a text-to-image image generator called Stable Diffusion. Despite how new this technology is, these types of programs work surprisingly well in interpreting and transforming written language to a visual one. According to Stable Diffusion’s website, every piece that it creates is wholly original, even with many of them looking like some generic No-Name Brand knockoffs of other more established artists’ works. Each image that I have created using this program technically belongs to me, so I now own the rights to not only the image above but all the ones featured in this writeup even though I haven’t invested very much time or effort into actually making any of them. 

Stranger still is the idea that if I were to continuously click “Generate Image” without screenshotting any of the text-to-image creations, then they will never exist or be seen by anyone else ever again. More often than not, this particular concept puts me in an anxious state where I feel compelled to save everything the program produces, whether what is on my screen is any good or not. Theoretically, the more keywords that I input into the software the more precise and detailed the resulting images should be. Except that this process usually results in the software creating a grotesque vision of what it believes is in my head. In other words, the program misinterprets the written language being used, creating visuals that seem to be straight out of a horror film. Hands and facial expressions are especially difficult for these text-to-image image generators to render accurately. Just take a look at the result below, generated by the keywords “they are all going to laugh at you”.

As you can see, spelling seems to be another issue, one that I will ignore for today.

There are also ethical issues surrounding the use of these types of AI image generators in creating nude images, promoting hate and violence, and/or featuring identifiable persons in any of the resulting pieces. Because Stable Diffusion is open source, the potential for users to modify the program down the line, making it possible for them to generate amoral, unethical, or illegal content, is certainly a genuine issue that needs to be considered. 

Additionally, some people have already expressed their concerns surrounding their current employment, believing that these programs may make jobs in the fields of advertising and illustration obsolete. Because I apparently own the rights to everything that I have been creating while using this software, it becomes very easy to produce visual art quickly and rather effortlessly in any style I want. Why would I hire someone to complete a costly piece of artwork when I am able to rip off other artists without fear of legal recourse?

So, is this the death knell of visual arts as we know it?   

I am not sure. But I don’t think so.

I am reminded of the conversations that were circulating around the band Radiohead circa 2007. At the time, the band was challenging the music industry by providing a “pay what you want” model for their then-newly released album In Rainbows. This was upsetting for many in the music industry because there were claims that Bit Torrent sites were devaluing music being produced and that maybe big-name artists like Radiohead could survive this model but that smaller, less popular ones likely could not. At the time, album sales were plummeting and the record industry was collectively concerned that with the advent of these new online platforms – and Radiohead’s bold album release strategy – music would become effectively worthless. And while the music industry as a whole has never really returned to the sales figures and profits from its heyday, musicians today are still able to make a living – albeit, one that is likely noticeably smaller than before. 

Similarly, if more established visual artists can continue to make a living creating art in a world where art can be made by AI, would less famous visual artists still be able to eke out a liveable wage? Where might visual arts belong in a world where original content can be produced for nearly free? 

Again, I am not entirely sure. 

At the moment, text-to-image image generators seem to be questioning the inherent value that we have previously put on making, consuming, and appreciating visual art. Similar to YouTube and Bit Torrent sites that provide many artists, filmmakers, and musicians’ works for free, we now live in an era where AI can simply replicate any artist’s style of artwork without ever having to credit or compensate them for their efforts. And this may lead to less work for fine artists, illustrators, graphic designers, and the like. Conversely, it may also open up new, previously unexplored revenue streams that haven’t previously been thought up or considered. And in this way, perhaps artists need to shift how they identify themselves.

Again, it feels too early to know how this will all play out. What I do know is that there is value in the process of making art, not just in the “products” being produced. Using your hands to manipulate various artistic media has been shown to have a therapeutic effect on the mind, calming and alleviating the stress and anxiety of those that put in the time to engage in the process. Dr. Brittany Harker Martin (2022) from the University of Calgary has recently researched this process at length, referring to this state as flow – what psychologist Mihaly Robert Csikszentmihalyi describes as a highly focused and productive state in which participants lose all sense of time and space. This state of mind has been shown to be healing and therapeutic for participants (Martin, 2022). These aspects of manipulating and engaging in these processes are missing when one engages with AI text-to-image generation; all that you are left with is the end product. 

Right now, the novelty of my newest toy is keeping me engaged. But for how long? This new medium has my attention but I feel that something is missing. And this leaves me with feelings of hope that text-to-image image generators will not eliminate or supplement the act of art-making for many. To be able to endlessly produce original content in a matter of minutes using any style and combination of words is difficult to wrap my head around. But for now, I don’t think it has the power to quash the human desire to connect and express ourselves creatively.

References

Martin, B. H., & Colp, S. M. (2022). Art Making Promotes Mental Health: A Solution for Schools That Time Forgot. Canadian Journal of Education/Revue Canadienne De l’éducation, 45(1), 156–183. https://doi.org/10.53967/cje-rce.v45i1.5017