Part 1: behind The work

The evolution of publicly available dedicated generative AI tools, from their initial release to the present, spans only a few years, and it's already challenging fundamental thinking about ethics, morality, and law. Generative AI faces strong opposition from the established art world itself, who frequently raise questions regarding its authenticity, appropriation, and provenance. Additionally, it’s widely perceived that images created with this technology are often generated basically autonomously with a click of a button, are all equally easy to create, and have no serious artistic merit. While I understand this sentiment, I’d argue this misconception is largely held by those who are unfamiliar with and misunderstand the skill of prompting. And this confusion is valid. AI definitely has the ability to automate, and yet still, its uncanny accuracy in replication can be bewildering, especially to novices. However, as an artist, I looked beyond its cheap trick and soon became more interested in pushing its limits. Like many interested users, I initially created thousands of Midjourney v5 images that were both intuitively easy to produce and yet completely mesmerizing. Eventually, I began to discover ways to use generative AI to create more unique images. As I delved deeper into various creative workflows, combining different software tools, and creating many thousands more images, I arrived at this art series. Not only does this collection of AI photography invite debate around objectification, judgment, and interpretation, the AI element demands that the viewer validate their definition of authentic artwork. Moreover, these images compel the viewer to contemplate that which opposes what we believe to be verifiable. They present a narrative that challenges our preconceived notions of what we collectively believe to be true. Regardless of personal preferences, these photorealistic AI-generated images affirm the plausibility of a new burgeoning reality, where empirical evidence will be considered a perspective and very little that we observe will be accepted at face value.

Generative AI software, even in its current state of development, can create perfectly realistic images in a wide-ranging variety of subject matter. Yet it can still randomly produce equally astonishing and absurdly inaccurate or erroneous results without recognizing its own errors. Initially, I sought to harness the software's power to achieve perfect renderings, but then I found its flaws more intriguing. It was at that point that the puzzle came into focus, and I decided on the final look for the collection, capitalizing on these idiosyncrasies. In essence, these images are the result of software anomalies. The strategy behind this body of work was to manage these behavioral anomalies to ensure consistency.

While platforms like Instagram are inundated with fantastical generative AI images–some reflecting true talent, many astonishingly beautiful images are created with minimal input by the user. And therein lies the unique dichotomy of generative AI: expert output is not at all dependent on one’s own expertise. It’s illusory. Generative AI is extremely capable and adept at enabling any user to achieve unparalleled results without any skill required. On a basic level, the artistic challenge has largely been removed. There can be perceived challenges, but the rewards still come easy. Once you know what to look for, the traces of its autonomy can be easily identified across a vast majority of images found online. Where the AI tools become most interesting is when you start to successfully circumvent its default proclivities. The reality is, the more you advance your knowledge of the software and attempt to control the outcome, the more exponentially frustrating the software can be to use. Luckily for the novice, the learning curve for some of these programs is practically flat and so a high degree of satisfaction can be achieved within seconds of using the app for the first time. Hence, most of the generative AI images online were heavily guided by the ease of the software’s automation.

Being able to master your creative tools is essential for achieving predictable results. And without predictive results, you’re simply guessing. Or hoping. I think in addition to its autonomy, a crucial reason for the lack of precise control with generative AI is because the tools are strictly interpretive, removing any intuition from the creative process almost entirely. If it doesn’t explicitly know what your intent is, it cannot intuit. And to make things more difficult, it often fails at interpreting. Sometimes, it takes instructions (prompts/tokens) quite literally or in unfamiliar ways, akin to a child throwing a ball when asked, with unforeseeable outcomes. Providing instructions to generative AI often feels equally confounding, with no clear rationale. As I just mentioned, typically, the more specific you want to be, the more challenging it becomes to achieve your desired output. But if you’re regularly satisfied with getting results that are not exactly what you envisioned, then it’s a win. However, if you want to achieve a higher average of precise results, that’s where the hard work comes in. That having been said, achieving exactly what you asked for (in 2023), is simply not possible with generative AI. There’s always an element of chance. It’s just how it works. These challenges will undoubtedly diminish as generative AI tools evolve and become more widely adopted. Nevertheless, the theme of control, or lack thereof, played a significant role in the creation of this collection. The level of resistance I encountered with these tools can only be truly appreciated through firsthand experience.

Since the beginning of time, greater value has been assigned to things perceived as rare. The rarer something is perceived to be, the more value it is given. This principle is especially true for art. Today, generative AI enables millions of images—ostensibly labeled as “art”—to be created daily by countless individuals who are not artists. How do you think the established art world, which places the highest value on rarity, would respond to this democratization of art?The answer is, they dismiss it altogether. While academia, art critics, and traditional collectors may eventually come to recognize the value in some AI-created artwork, rarity will remain central to the art world. And generally speaking , the millions of images generated daily on platforms like Discord do not constitute rare art. An early example of the mainstream art world accepting and legitimizing AI-based artwork as a new genre is the MoMA’s acquisition of Refik Anadol’s Unsupervised. However, it's important to note that Anadol is an established artist, and his artwork is wholly original. While it was created using AI, his work bears little resemblance to the vast repository of images on Discord.

Some believe that the advent of AI is poised to be the most disruptive human innovation in documented history. Whether this development is for better or worse will likely remain a matter of debate for some time. The role AI will eventually play in the world of traditional art will undoubtedly take time to develop as well. Beyond the obvious aesthetics, I hope my work will make a meaningful contribution to the conversation.

Part 2: focused attention

In the following paragraphs, I will address criticisms that these images only serve to proliferate the objectification of women’s bodies. Firstly, my decision to explore the female form in this way is nuanced. But given the photorealism, beautiful aesthetics, and highly stylized deformities, the negative interpretation is unsurprising. Ironically, the work is intended to challenge complex cultural, philosophical, and emotional themes that, in fact, belie the very argument. As the female nude has been a central subject in art for centuries, I too found this subject matter an ideal vehicle for showcasing generative AI’s “mind-blowing” ability to replicate human anatomy. As a result, I was able to explore human qualities such as vulnerability, power, and sexuality in new ways. The fact is, not only does the female form here serve as an important aesthetic element in the work, I’ll argue it was essential for achieving the intended effect. Nonetheless, regardless of any artistic merit I proclaim, I believe accusations of this nature deserve a dedicated response.

”There is a thing about beauty. Beauty is always associated with the male fantasy of what the female body is. I don’t think there is anything wrong with beauty. It’s just what women think is beautiful can be different.” – Jenny Saville

I realized from the start that some viewers would respond negatively to these images. I don’t want to seem flippant, but for some, that’s a reasonable response. These images are intentionally unapologetic in their frankness and provocation. I obviously don’t wish to be hated or accused of being malicious, but sometimes art‘s purpose is to elicit an intense response.

Lastly, I feel an important distinction to make here is that nudity was not the impetus for this work. As with a lot of creative projects, it evolved from an extended period of exploration followed by a lot of trial and error. As a point of interest, nudity is not a recurring theme in my work. As for my rationale to use the nude female form in this body of work, I hope the circumstances behind it and the significance of it becomes more apparent as I delve further into the topic.

As I mentioned already, concerns that these images objectify women's bodies have been brought to my attention. Though I naturally anticipated that this could be a perspective taken by some viewers, suffice it to say, it is not my intent. While I have no choice but to accept the liability, I’m frankly dismayed by the misdirected attention. For the moment, consider this point to ponder. Even as viewers are aware that these images are AI-generated, this knowledge has done little to counteract the visceral response that the images provoke. This could simply be attributed to a perceived salacious and gratuitous display of nude women. But I would propose the underlying trigger is that these images personify the insidious effects of AI technology in a manner that only such an image can effectively communicate. In contrast, I'd confidently argue that a clothed figure would not have had the same impact. Similarly, I contend that nude male torsos would not have achieved the intended effect either.

My approach to working with AI-generated nudity was no different than working with nude models in the real world. They’re undoubtedly not equal, but the sensibilities required are no different. Given the obvious delicate nature of the work, whether working with AI-generated nudity or working with a live nude model posing for a fine art photo session, my role and responsibilities as an artist require the same tact. In one scenario, working with the live model requires decorum, compassion, and empathy. While in the second scenario, although the live model is removed, my humanity and personal ethics remain the same. My underlying principles that enable me to achieve success under the first scenario has no need to change for the second. In either instance, a principled understanding of the framework is integral to its success. Although the female nude form is boldly displayed here in a manner that some people may find less than aesthetically pleasing, maintaining my professional integrity in the pursuit of the highest possible degree of artistic truth was my main focus.

There’s also a technical facet to this debate. On the topic of female nudity specifically, it is in my professional opinion that the specialize AI software I used objectively favors females (as defined by the AI) over males. In other words, the AI generates photorealistic images of the female form with more realistic accuracy than that of the male form. In my opinion, the female skin tones and details are rendered more realistically. Even light tends to illuminate the female form more precisely than their male counterpart. My theory behind this is that the mass number of trained AI datasets are skewed toward images of females. Therefore, to put it bluntly, the AI knows what women look like more than it does men. If my theory is true, it arguably points to more serious biases hidden by software developers. However, some AI experts and founding software developers have, in general terms, contradicted this theory. But I continue to maintain, that for whatever reasons, the generative AI software I used creates more realistic looking images of the female form. And since my goal was to create the most realistic images I could, I chose the female form as my subject.

There’s no arguing from me that the nudity is the artwork. But as I’ve mentioned, nudity was not the starting point for this project. Indeed, this project largely came about due to serendipity. One discovery led to the next, which led to the next. But with the highly unpredictable nature of generative AI, along with the current limited and arcane software available, exploration is a primary component. It’s practically written into the code. And with that in mind, I cannot stress enough the importance the AI software played in this endeavor. The AI software set the parameters, provided the inspiration, and ultimately determined the nude subject matter. It goes without saying that my contribution to this series is the work itself, but it's important to acknowledge that without the AI software, these images would not exist.

Finally, I'll bring this discourse to a close with a few remaining thoughts. I’ve proven to myself here, any sincere effort to justify the creation of images featuring a female nude form, gratuitous or not, is particularly onerous. So, contrary to any attempts at clarifying what my intentions were, as much as I've tried, I concluded before I began that no explanation defending the work would suffice. So, in light of that, I will simply offer a common definition of art: something that is created with imagination and skill and that is beautiful or that expresses important ideas or feelings.

All things considered, I believe that’s exactly what I did.