On AI Art

The admission that I have used generative AI in any portion of the process of creating my books is something that will, in our present cultural climate, cause a fair number of people to drown in presumption about what type of person I am and what type of “work” I’ve actually invested. I cannot combat presumption and I cannot prove what will not be believed. What I can do is talk about my perspective and my process. For those who are inclined to listen, you’ll find an open invitation to judge for yourself whether my efforts are worthy of your attention.

It would be ignorant to claim the inclusion of generative AI in my process brings no implications to bear. I know it brings many implications. I have a perspective on the usefulness of this tool and I have the open-mindedness to recognize some of my thoughts can easily be construed as hypocrisy. Like so many others, I do tend to view the effort I’ve invested as inherently valuable. I do also frequently apply this value to the end product as though it is the determining factor. I do this even while claiming the effort itself should never be it. I am guilty of looking down on what I deem to be “lesser” efforts while justifying what I deem to be “sufficient” efforts.

Beyond simply patting myself on the back and rehearsing mental reassurances, though, there is some honest thought to apply to the present situation. Changes - particularly changes with the kind of sweeping impact that large language models and generative AI impose - affect people in a variety of ways. Some ways are pleasant. Some ways are unpleasant. Some ways are beneficial. Some ways are harmful. I’ve taken some pages here to proffer some musings on these issues inasmuch as they touch on the work I’ve presented.

The Problem of AI 

There are many arguments against the use and implementation of so-called "artificial intelligence." While I would likely start the discussion by recognizing that "AI" is itself a misnomer, the broader discussion about the impact of its existence is still important. I would argue that the complex computational matrices being employed are not actually intelligent but are instead simply robust (perhaps in unexpected or even unpredicted ways). The larger concerns surrounding AI - more specifically generative AI - cover more topics than I will address here, but I have compiled a short list in an attempt to articulate what are perceived as the problems with AI.

Contention #1 : Generative AI cannot understand requests and therefore cannot employ any form of vision or intention in the creation of its outputs. 
I agree that generative AI is not a useful tool for revelatory self-expression. To whatever degree an individual presents an epiphany, the involvement of generative AI in the development of that concept would serve only to homogenize the idea with existing established concepts previously used to train the AI model. Only an exceptionally powerful and unassailable vision would be impervious to such an influence. This begs the question of why generative AI would be used in such development at all. The role of generative AI in this scenario is to bring the new concept in line with existing ideas.

Contention #2 : Generative AI cannot improve upon the existing depth or veracity of artistic expression and it furthermore cannot correct inaccuracies and biases which exist within the information used to train it. 
I also generally agree with this contention. It might be important to understand the vastness of the training data in order to contextualize this point properly, but even in the absence of that it is obvious that only the ideas which have existed and been expressed can be replicated by the mechanism used to collate such expressions. New expressions or new explorations into the vagaries of reality cannot be generated by this system. Whatever output is spawned by the system is definitionally a restatement or, at best, refinement. Because of this limitation, I think it is also fair to presume that the biases which already exist in the utilized data will be perpetuated in its outputs. 

Contention #3 : Generative AI hallucinates outputs that do not exist in reality.
This is not a conclusion I would have come to on my own. This point contradicts, to some extent, the prior point about the limitations of the expressive capacity of genAI. This phenomenon is observable, though, in actual practice. In fact, genAI frequently "imagines" outputs which do not seem to match the training data it is working from.

To my understanding, genAI is a system which predicts what the most likely pixel adjacent to the prior pixel would be. It does this based on the parameters of the data set used to train it. This is a greatly simplified explanation of what is happening, but in the context of this understanding of genAI outputs it becomes logical that some of these outputs would veer away from the constraints of the natural world. Some artists, after all, take liberties with the fabric of reality in order to express something beyond simply the replication of what is.

The functional problem created by this contention is that genAI output can't be "trusted" like something "real" could be (e.g. stock photographs). If using a genAI output as a reference, it is impossible to know whether the shadows or shading of the output follows reality since it is simply an interpretation of other creations.

Contention #4 : Generative AI imitates the style of copyrighted materials and imitating the style of copyrighted materials is bad.
This contention, while popular, is utter nonsense in my opinion. I believe that almost every artist that has ever existed has built upon the influences of those who preceded them. In fact, even if one imagined the very first man, Adam, to have been an artist, he also could only have followed after the artistry of the creator himself. The idea that replicating the style of another artist is in some way reprehensible is itself a deplorable form of gatekeeping. 

We are, all of us, influenced by the ideas and expressions of those who have gone before us. It is our responsibility to have some humility and recognize that we are not ourselves creator gods. We are merely human, with the ability to learn and grow like all other humans do. When we create, we actually innovate. When we express, we actually compile. When we explore, we actually only observe what has already been seen. To invent a restriction that imagines we could do anything without the influence of others is to falsely imagine it is possible to do so. It is not.

In this sense, the capacity for genAI to apply the style of talented artists to the raw verbal concepts we might present to it is simply a triumph of ingenious automation. This isn't a bad thing. It streamlines an arduous process that, for some, accomplishes much for no reason. Whether or not genAI is a tool which will benefit any specific effort is actually a matter of personal determination. Is the growth wrought from effort needed more, or is the functional output needed more? Only the people involved in each situation can know.

Contention #5 : Generative AI puts people out of work.
This is undoubtedly true. I would take exception to the allocation of responsibility, though. History is littered with examples of newer, better, and more efficient methodologies emerging within markets and workplaces. I do not think the rise of genAI is substantively different from any of these other shifts. Sometimes the way tasks are accomplished changes to accommodate a new approach. Sometimes the new approaches are robust enough that they affect far more than simply the task which inspired its inception. Clever people apply clever ideas in clever ways. As we peer back on other such shifts in how work is handled, we tend to view them favorably. We have the benefit of seeing how the change has benefitted all without the burden of worrying about the people who were forced to reconcile the change itself. 

Yes, genAI puts people out of work. It eliminates work. It changes the types of work needed. None of this is an inherently bad thing, given that appropriate concern and compassion is afforded to those whose entire lives and livelihoods are compromised as a result.

Contention #6 : Generative AI harms the environment. 
The word "harm" in this contention attempts to circumvent the process of analyzing the new impacts of these systems. Generative AI impacts the environment, certainly. There is an enormous new demand for compute resources in order to drive the computations of generative AI models. This has driven a rapid expansion of data centers which require incredible amounts of electricity to function. Depending on how this electricity is generated, this demand will have varying degrees of impact on the world around us. While the degree of the impact may be adjustable, the fact of it is certain. In light of this, anyone who views generative AI as an evil or a detriment to the state of humanity would naturally conclude this impact is equally terrible or unworthy. Therefore, whether generative AI harms the environment is really a discussion centering on whether its impact upon our resources is worthwhile. Is there any benefit derived from generative AI which offsets and justifies this expense?

Contention #7 : Generative AI inhibits human learning.

This is the idea that convenience begets laziness and laxity. There is observable truth to this sentiment, as many who incorporate automations lose the core skills which have been automated. The example of a student who relies wholly on a calculator becoming unable to perform calculations on their own springs readily to mind. A skill loss absolutely occurs when processes become automated and streamlined. 

This has a dual application, I think. In the case of an artist who aspires to hone and improve their artistic ability, I think genAI could certainly represent a net loss. In this case it is important to encompass the very skills the algorithm is replicating. Without the practice of working on and with these skills there is a very real danger that the artist will lose fundamental competence and may even find themselves handicapped or incapable in the face of their own inspiration. However, in the case of a product deliverable, I think genAI represents a massive overall gain. The creation of a product does not necessarily pander to the extension of personal skills. In fact, many such efforts only support such investments begrudgingly in order to afford the most efficient available path to resolution. In this example, genAI will embody a massive leap forward in efficiency. The lifecycle of production benefits enormously from eliminating unnecessary investments like developing individual skill sets. This is often misconstrued as contrarian and perpendicular to human learning where it should be recognized as parallel. GenAI is, in this scenario,  a work multiplier which allows more to be accomplished with less. This is a benefit to everyone involved.

The Problem of Time 

In my examination of literature there is revealed an event horizon breached by all who strain breath into thought. It is this: that our lives are limited; our time a finite resource. Eventually, we all must contend with the reality of capacity. I, like anyone, am capable of many things. In my youth, I had no concept of how many things and I passed the consideration of limitation off to the infinity of undefined potential. As I have expended my life, though, it has become increasingly obvious that my undefined potential has a definitive ending. These are fancy words to express that I can only accomplish so many things with my life. It is my responsibility to prioritize my efforts according to my increasing limitations. 

“Prioritize” is a hideous word. It is all restriction and measurement with no space for hope in illimitable creativity. Yet, it is also the fathom of existence; necessary to the core of our sin. It is essential to arrange all that I am according to the hedge of my life. As much as I might love my capacity to become an actual artist and divest my efforts wildly at such a realization, I also understand with accelerating clarity the cost of doing so. To become this thing will be at the expense of some other thing. The moments of my vitality are draining away. So, while the unbounded enthusiasm for all edification is life-affirming, the liability of mortality remains ever-present. I can, and must, choose wisely.

As I conceptualized the story I desire to tell, I allowed my mind to run free. Wherever it alighted became a concept worthy of exploration. In this heady euphoria I planned to write hundreds of books, thinking I had nothing but years to fill. I wrote lists, organized acts, and build a grand rendition. My only trepidation was for the style overlaid on all this substance. Awash in self-satisfaction, I daydreamed years away. I did not do so in futility. I simply focused my priority elsewhere - outside of narration. 

I am not now dissatisfied with the course of my existence - far from it. I am grateful, if anything, to be where I am today. Coming to this home once again, I am cradled by the wealth of concept awaiting renewed attentions. I have to decide now, though, what can be worthy of travail. It cannot be hundreds, I think. There are not enough years available. I don’t want to give any audience the anguish of an unfinished tale. At least, not if I can help it. 

This is the crux of my presentation about genAI. There is only so much I can do. In my heart, I want to be a writer and I want to write the story I’ve envisioned. I wouldn’t give that away to anything. Not to learning some ancillary skill. Not to the mindless regurgitation of the common. So, yes, I employ genAI. It helps me create the images I believe I need without investing the time I believe I no longer have. It does not help me write, though it could. I don’t want it to. I am too jealous of my recitation. Telling the story is the thing I desire. 

The Problem of Production 

One of the fundamental conflicts of writing, in my opinion, is focus. There is a tug-of-war between concept and execution. When an idea is first encountered, it is pure and beautiful. The very first task is to capture a part of that beautiful purity by considering it. Capture it by ingesting it. Some would say to do so by meditating on it. I am certain the specifics differ for each artist, but I must to some extent embody the abstraction to give it a seizable substance. Until I can grasp the quintessence of the caprice, I can neither share nor present it. 

Once I have savored whatever portion of the purity I can master, I am now able to move from conception to execution. This is where we can commune together. Where knowledge is entertained. This is where the real work of the process exists - in bridging the consensual expectation to the peculiarity of the new. Which is simply to say that just as much as people desire the titillation of intellect, they equally require the strictures of expectation. Very few people are willing to engage with an idea that isn’t presented in a familiar or palatable manner.

Therefore, it isn’t enough to explore an idea. It isn’t enough to delight in cogitation. It isn’t even enough to caress proposition. The pure and beautiful idea must be translated into a form relevant to its audience. Without the work of translation, the idea itself retains no pertinence.  It is lost to the insignificance of the individual. There is no version of conception that meaningfully serves which does not involve the discordance of connection.

This begs reconciliation with the audience, because not every notion serves every audience. Not every thought should serve every audience. So, I as the writer must first discern the breadth of my audience. Who am I writing for and who am I not writing for? How much of my audience am I willing to stretch to reach? How much of the idea should be sacrificed to this effort? Because it is a certainty that some of the purity of the idea will be sacrificed. Also, what capacity do I have to negotiate this sacrifice? Some writers are better in the translation than others. One writer might experience profundity on an inexpressible level while another might rephrase the bromide into a revelation. There isn’t a standard judgment for either one. What I want to be and what I can be are matters of skill, desire, and intention. Any one of these three things can become a force which overwhelms the others to my detriment. 

I’ve used vague and inclusive wording to convey the struggle, but it is actually a simple thing. I am telling a story. Who am I telling it to? What parts of the story am I willing to sacrifice to make sure it reaches my audience? What parts of the story am I not willing to sacrifice? And, finally, how do I accomplish this work? This, I think, is the problem of production.

I write all of this as a person who has spent the better portion of 35 years wrestling for answers to these questions without grasping any functional resolution. I think that probably puts me farther from the archetype of an effective writer than I’d like to imagine. I have lived in a tradition of self-deprecation. I would practice this to avoid pride and embarrassment, but I must also grapple with the possibility that I’ve mostly just been sabotaging my own efforts. 

Again, all of this reasoning might be overdone. I still have to struggle to determine what art I can produce that is good enough. Good enough to share. Good enough to accept. When you’ve read my story and weighed it against the amount of thought I’ve invested into its presentation you might detect an imbalance. I certainly do.

When I create the product that encapsulates all of the ideas and struggles I’ve suffused into my writing, I want it to convey the interest I expect my labor to generate. This means some aspects of the product are going to have some standards. Even if my standards are poor, I still have them and I still want to meet them. This is a romantic way to argue the images I draw by hand, unassisted, are not good enough in my own opinion. When I approach the task of creating a book cover, I have certain expectations that I lack the skill to meet. So, the heart of the problem with production is the question: how will I make it good enough?

The Solution I've Used 

As a person who isn't trying to be an artist in the traditional sense - or who feels particularly interested in gaining the attached accolades - I've looked at the artwork I want as something of a hurdle to overcome or a commodity to obtain. This isn't to pretend that I am not proud of the work I've done in generating the images I use for my book covers and cards. I am immensely proud of what I've accomplished - and particularly so because I have done this with the enormous handicap of lacking any real training or practice in the knowledge and techniques I needed.

I've read ad nauseam that a person like me is not an artist and the images created by a process like mine are not artwork. I have made my peace with this assertion. As I look out into the world and observe any of the many talented artists who have devoted themselves to honing their craft, it is easy to honor these individuals as a separate class of talent and capability from what I possess. Any attempt to judge myself against them would result in the harshest critical reprisal. Mostly, I trace and color from reference which simply isn't the same endeavor.

But I am not hung up on being labeled an artist, nor do I covet recognition for some presupposed greatness. I am not great. What I am, and always have been, is a person who is rather adept at solving problems. When I started writing, I did so out of passion and interest. I wanted to tell a story and I spend most of my free mental energy imagining the details of the story I am trying to tell. I would love to be recognized as a writer, to be enjoyed as a writer, or even to be respected as a writer. These things sound incredibly valuable to me. I've pieced together my thoughts and plotted a course for setting these thoughts to paper and this is exactly where my interest in the process of writing meets its end. In writing and writing alone.

I found that there is so much more to the practice of writing than simply putting ideas onto a page, though. I can confess with very little reservation that every single other thing that I've been told is "necessary" or "essential" to the practice amounts to little more than an inconvenience or annoyance. These are just problems that it seemed like I must solve. Therefore, when I needed some kind of artwork to put on the cover of my book, my first solution was to position stark lettering in the middle of a black field. I thought that looked just fine. I was told otherwise.

Then I thought I would create a simple sketch which I, again, perceived to be rather fetching. Again, I was told this was insufficient to the task. I would not succeed in garnering the interest of any potential reader with such a meager effort. 

Then I was instructed that what I really needed to do was hire an artist to create a cover for me. I found this irksome for several reasons. First, I smarted at the accusation that the things I generated were somehow not good enough for my potential readers. Second, I lamented the very concept of enticing anyone with anything other than the purity of my own writing. Third, I lacked the resources to adequately fund such a venture (and I do not begrudge artists their due - I simply lack the due to bestow). Fourth, nothing about the proposed solution guaranteed any sort of success anyway.

I found myself paralyzed by these frustrations. It occurred to me that I might train myself in the discipline of creating a piece of artwork that would be deemed sufficient, but this seemed inordinately time consuming. With the advent of internet services I pondered the possibility of commissioning an artist cheaply, but discovered this would not grant the kind of rights I would need to actually utilize the end result. I thought I might convince a friend or acquaintance to generate the art as a gesture of goodwill, but then I became reticent to take advantage of someone in such a crass and self-serving manner.

Why do I belabor this point? Partially because it was so laborious for me. So many years of frustration and dissatisfaction have been invested into these problems. Perhaps more importantly, though, is because I want to endeavor to help you, the reader, understand the context in which I finally seized upon a workable solution. A solution that I found myself quite happy to grasp. 

While the world around me obsesses over spewing hate at generative AI, my experience is quite different. Generative AI was a breath of fresh air. For a relatively nominal and reasonable fee, I could sling prompts at the algorithm until it spat something back at me that approximated my vision. Taking this output as my baseline, I could quickly and easily trace and modify the image to create what I deem to be an exceptional end result with minimal hassle and time investment. This, to me, was a breath of fresh air.

I acknowledge that for many readers, this prior paragraph may be ugly and offensive. That I could create an image which I had developed neither the skill nor the proficiency to generate though traditional means appears to be raw sacrilege. Not to mention all of the suffering artists who won't be employed by my need for image assets. And generative AI "stole" art to train from (though I believe the courts have disagreed with this contention). And generative AI just stitches other people's work together without bothering to learn their techniques. And generative AI is putting us all out of jobs. And generative AI is the tool of the antichrist. And generative AI will destroy modern society as it brings on the Armageddon foretold from ages past.

How could I?

Despite my hyperbole, I mean the "how could I?" in earnest. This is a question worth wrestling with. How I did was in a few easy steps. First, I checked to see if the generative AI would leave me with any copyright issues to resolve. It would not. Then, I checked to see if I was capable of altering the generative output to suit my vision. I was. Then, I checked to see if this process could be accomplished within a reasonable amount of time. It could. These needs verified, I pushed forward and solved more of the issues which had long hampered my ability to present my writing to the outside world. 

It is likely any reader of this essay can surmise that I've also gained a fair amount of trepidation toward the process I've used in the face of widespread social pressure to its contrary. I understand that some people will not accept my work simply because AI was involved in any part at any point. It doesn't really matter what I say or how much I explain. Nothing will be good enough. I think that's sad, and perhaps a little bit hurtful. It's not really mine to control, though.

For the rest of my potential audience, the best I could think of was to simply share the details of my process so that you would not feel like you've been duped or deceived. To that end, here is a description of the step-by-step process I use to create the artwork you've seen on my book covers, within their pages, and (possibly) in my card game:

1) Imagine what the image should look like.
2) Interact with a generative AI tool and attempt to describe what I've imagined.
3) Render outputs and make adjustments until something approximates my vision.
4) Examine the chosen output(s) to determine what needs to change.
5) Trace the output and establish the lines I want while making adjustments as needed.
6) Bring the sketch to others to see what they think of it and collect any further changes.
7) Revise the sketch and repeat 6&7 as much as needed.
8) Create a new layer to add in what I call "hard shadows."
9) Create a new layer to add the base colors for the image.
10) Create a new layer to add the shading (using the original output for reference).

That's it. Aside from the composition being bound to whatever vision I have for the piece, you can make your own judgments as to what kind of skill or effort is required to get through steps 1-10. I promise it is not as much skill as an actual artist requires to make whatever art they are creating. What I am doing is generating a product to fill a need, and I am trying to be careful not to claim it is art. What constitutes art is a much deeper discussion that is probably best left to the artists.