The Process is the Cake

This essay is the start of what I imagine to be a lifelong reckoning with the ethics of AI. In the context of art, the topic branches in several directions: the relationship between technology and art, the integrity of creative process and what gets lost when it’s skipped, disclosure when AI is involved, the labor and consent issues around training data, energy & environmental concerns, and the market compression that threatens people’s financial stability. I understand that any one of these branches alone can be enough to turn a person completely off from generative AI. Training data and energy use are two stand outs in mainstream conversation right now. But I wanted to start with the less technical side: what does generative AI mean to art, how could it change the artistic landscape, and what could process integrity look like?

Defining Art

I see art as the process of expressing a feeling, truth, or vision through applied creativity over time. That word — process — is critical. Art is about presence, about putting yourself in touch with something that opens you to vulnerability, feeling, and surprise. Some artists are futurists and prophets, some are feelers and healers, some are creators, some are destroyers. There are a billion ways to do it, with new ones emerging every moment.

Some take a more material definition. A painting is “art.” A recorded song is “art.” For me, these are examples of art’s outputs, not art itself — artifacts of the artistic process that, in pure isolation, are not the thing. I find this distinction especially helpful when thinking about AI. By this framing, a direct model output isn’t yet art on its own. It’s the result of fusing together countless hours of rendered process from creators, with a prompt pulling out the forms requested by the prompter. It only becomes art when it’s folded into a human process of meaning-making.

Our current norms around art, particularly intellectual property, don’t line up with this interpretation. But I’ve long felt that art and enterprise intersecting is an exceptional phenomenon and not a sustainable one: at some point in that relationship, art ceases to exist. If you can make money from your art, great. Keep going. But expecting transactional economic systems to reliably serve art has always been a fragile arrangement. Businesses are built to increase their bottom line; artists are built to feel and express. They may cross over, but that crossing is exceptional, because these two impulses are inherently misaligned.

This matters for the AI conversation because it helps separate “AI threatens my current job” from “AI threatens my art.” As technology advances, the former is nearly inevitable, and that can be a painful loss. But the latter being true demands a more fragile perspective on art than what I’ve defined above.

Using Technology as an Artist

Technology and art have a timeless relationship. Tuned instruments were once a novel technology that enabled sounds previously reserved for the human voice. A guitar can produce a note more pitch-perfect than most humans can sing, and it can play chords — technically possible for one voice but incredibly difficult, effectively requiring multiple singers. In this way, instruments could be considered a “cheat code” for sounds humans would have otherwise produced with their bodies.

I wonder if the creator of the first musical instrument drew ire from their community, or were they celebrated for enabling new forms of music? I wouldn’t be surprised if it was both. Any time someone challenges a norm, they tend to be celebrated and disliked. Change can be both exciting and discomforting.

Fast forward to electronics, and you see the same pattern with hip-hop. Taken in terms of time spent listening, creating, and being connected to the scene, my deepest roots in music are hip-hop. While I was born into a white, hip-hop-distant family, discovering rap as a child changed my life. It opened me up to new people, ways of thinking, cultural understandings, and feelings I couldn’t access through other music. Hip-hop would have had a hard time existing without electronics. Turntables. Sample machines. MIDI. Home studios. Boomboxes. These were unlocks that enabled generations of inner-city people to express themselves. To this day, some people talk down on hip-hop because it breaks so many rules of what came before it. It is the avant-garde movement of the late 20th and early 21st century — the dissolver of genres. It doesn’t ask permission. Its use of technology is central to that.

So What About AI?

AI in the music space is particularly complicated for me. Music artists spend years honing their voice, pen, timing, and production capacities. And yet, with tools like Suno, in about 30 seconds you can create a song that is, on the surface, technically brilliant. It almost hurts to experience — a machine spitting out something eons beyond your ability in terms of pitch perfection, timing, and production fidelity.

A small anecdote brings this home, but also offers a glimpse at the upside. I recently released a new song [humans] . I produced the instrumental, wrote the lyrics, recorded the vocals, came up with some effects layers, and paid to have someone mix it professionally. I like how the song turned out. It’s imperfect in many ways, but I felt comfortable enough to share it. Recently I decided to upload it to Suno and try the “cover” feature. I’m a huge fan of the (late) Beatles, so I asked Suno to produce a new version using a handful of sonic keywords from late 60s / early 70s music in that vein.

In seconds, Suno produced this [humans ai] . It shocked me. I listened back to my original and cringed. How did I miss this note? Why is my arrangement so boring? Why is the timing on my guitar so imperfect?

But then a curious thing happened. I found myself enjoying the song again through this new machine-generated version. Most artists will agree that after you’ve heard something a million times, it’s hard to feel the same way about it. Hearing the song from a different slant brought life back to it. It made me realize that I love the lyrics. I love what the song stands for. I love the key I ended up in. I even like the melodies and harmonies I chose, which the AI made some use of in its reprise.

And now I’ve been working on an acoustic version of the song — not directly using AI, but inspired by the new life that listening to a different rendition provided. I find myself practicing the notes, timing, and harmonies I wasn’t satisfied with the first time. Almost paradoxically, some AI-generated “slop” has led me deeper into the process of a song I’d largely felt done with.

That experience clarified something. The AI version was useful because it gave me a new angle on something I’d made, stripped of my own self-criticism and fatigue. It reminded me why I wrote the song in the first place. I make art to feel, envision, connect, and reflect. When I hold onto that, there is no world where the machine outdoes me, because those elements are innately aspects of the experience of being. No output can take that from you. You pay that cost yourself if you choose to skip the process.

Where I draw the line: I’ll never release the AI rendition as something “I made.” The artists in the training data, including my beloved Beatles, toiled for cumulative centuries to reach the sonic and rhythmic elegance required for Suno to approximate what “good” is. Suno applied those hard-earned qualities to the inputs of my song. I feel uncomfortable claiming an output made from that, while my own process is quite far from being able to reach it directly.

The Slop Problem Isn’t New

One thing I keep coming back to is that there has always been “slop.” Humans have been making the equivalent for quite some time. I find that a significant amount of what streams on major platforms is algorithmic, unoriginal, mind-numbing content, a.k.a. “slop”. It is the processed food of the mass media era. And yet, there are more artists than ever creating deeply meaningful work. This is a product of accessibility. As it gets cheaper to create something, more people will do so. And people are unique.

Due to conditioning, societal and economic pressure, many people aren’t oriented to take on art in the esoterically meaningful way I’ve described. When you’re stressed and need money, that kind of engagement often doesn’t get to occupy your focus. As humans are offered tools to make artistic outputs more easily, some will use them to solve external problems. Long before AI, I saw this playing out in hip-hop — artists focused on “making it out.” And yet, hip-hop is also full of people who carried that same hunger and still made something undeniably meaningful along the way. Roses can indeed grow from concrete. Scarcity makes process harder to prioritize, but it doesn’t make it impossible.

Inevitably, though, as we strip away inner process in favor of external reward, art becomes less present. The person skipping the process isn’t just producing something hollow for the audience — they’re robbing themselves. The muck of creation, the failed takes, the ugly drafts, the moments where you have no idea what to do next — that’s where the feeling lives. When you hand too much of that to the machine, the output might look like art, but you didn’t get to go through the door that making it would have opened in you.

If AI can produce polished outputs in seconds, the market for “skilled output production” will compress. The artists most exposed are those whose livelihood depends on producing things that look or sound impressive rather than on the depth of their process, taste, and experience. The artists most insulated are those oriented around something the machine fundamentally cannot do: be present, feel, and make meaning from lived experience.

There will be artists who use AI to do extraordinary things previously out of their reach. There will also be newcomers who finally feel enough traction with creation because of AI that they can finish something — which, paradoxically, could bring them deeper into process. Imagine a young kid who has passion for the pen but no recording studio, no musical friends, no equipment. They pump their lyrics into a model and hear what their song could sound like. That’s potent, and could be exactly the encouragement they need to go deeper. While art isn’t about its outputs, the validation one gets from them — even the machine kind — can be fuel that carries you through the early stages.

Draw Your Lines, Own Your Truth

What I’ve come to so far is this: do what you will as an artist. You don’t need permission to use the tools you want, including AI. But be honest about it. Don’t create a song purely with AI and present it as if you toiled with a mic and Ableton for countless hours. That’s inauthentic. People will feel it. And what happens when you want to do it live?

If you use AI heavily, remind yourself of your intentions. If your goal is technically impressive, marketable content, maybe heavy AI use is a cost-efficient path. But if your goal is to feel, be witnessed, grow, and enable others to do the same, consider the value of your imperfection — of being totally raw and completely you, missed notes and all. Most people will land somewhere between these two perspectives.

I will make use of AI for parts of my process, as I already have. As much as I want to feel, I also like pushing the boundaries of what’s possible. I’m still a work in progress regarding perfectionism, and I don’t always have the time or funds to do all my process manually or with other humans. That’ll turn some people off, but so be it.

One clear line for me right now is avoiding AI voice changers or vocal generators in the music I share. Two reasons. First, the human voice is one of the closest channels we have to our uniqueness — everyone has a distinct timbre, and I want mine maintained, even when it makes me cringe. Second, and related, I’m not satisfied with my vocal control yet. If I lean on AI here, it becomes a crutch, and I probably won’t reach the level I want for myself.

Writing occupies a similar space for me. I’ve been using AI to help grade / review my work (including this piece), automate mundane formatting, and help with polish & rephrasing. But I want to be a better writer, so I’m doing my best to not lean on it too heavily. Much like the sonic voice, our linguistic voice holds something unique. Beyond that, writing for human comprehension, bias, and empathy is difficult. With all the training data in the world, the frontier models are still terrible at it compared to a good writer. But I do wonder how it might help a good writer through some of the process stickiness that comes with the space.

Conversely, I’ve been exploring using AI more liberally to generate visual elements. I do a lot of manual process with visuals, though I’ve had people claim my work is “all AI”. In my teens and 20s, I committed thousands of hours to visual work, especially on the editing side (a few examples of my past work) . There was a time in my life when I knew Canon cameras, Premiere, and After Effects like the back of my hand. Personally, I find tools like Midjourney liberating. Sometimes for direct use, but also as a concept tool. With that said, looking back at a few of my visuals, there are segments where I wish I’d gone deeper on traditional processes. I’m still learning to strike a balance that feels good to me.

Everyone will draw their own lines. This has always been true to an extent. There are people who feel even touching auto-tune excludes you from being an artist. But I’ve experienced deep emotions both listening to music that uses auto-tune and creating it. Applied with taste, auto-tune can produce sounds the voice can’t do alone. Drawing your lines in a fast-moving world is an iterative process.

You won’t please everyone and you don’t need consensus to create honestly. Just remember the cost of giving up parts of your process to the machine. Make sure it aligns with what you want from your art. Honor the value of feeling things. Of being uniquely you. Remember that whatever you get externally from your art should be looked at as the icing on top.

The process is the cake.