AI as a tool for artists: what I’ve changed my mind about
There’s no ethical way to use AI yet, and the law will show up too late to matter. It’s still worth learning, as long as you treat it as a tool, not a replacement.
I wrote the original version of this post a while ago. Back then, I didn’t have a clear picture of what AI was going to do to this industry, or to others. My thinking has changed a lot since, so this is a full rebuild, not a touch-up. Five things stand out to me now, and I want to walk through each one honestly.
There’s no ethical way to use AI as a tool for artists right now
As of right now, there’s no fully ethical way to use AI as a tool for artists. Every major model learned, at least partly, from material nobody cleared the rights to. In fact, that’s not a minor technicality. It’s the foundation the whole technology sits on. No amount of careful prompting or thoughtful use erases where the training data came from.
The law is already behind, and it’s not catching up soon
Regulation always trails technology. That’s not new. But with AI specifically, the gap already feels too wide to close in time to matter. A real legal framework covering training data, copyright, and attribution is still years away, if it comes at all. By the time it exists, the current generation of models will already be the foundation everything else builds on. Waiting for the legal picture to clear up before deciding how to relate to this feels like waiting too long.
AI as a tool for artists: why it’s worth learning fast
Even with all of that unresolved, learning AI as a tool for artists is still worth doing. The professional landscape around it is changing faster than almost anything else I’ve watched in this industry. Artists and other professionals with a strong 2D or 3D background can do something genuinely new here. Once AI enters their process as a tool, not a replacement, the work gets more impressive, not more generic. The same shift is already visible in medicine and in tech. In both fields, it’s the specialists with deep domain knowledge getting the most out of these tools, not beginners.
If you’re an artist, even trying AI as a tool for artists on admin work is worth it. It frees up time for the part that’s genuinely fun, making the work itself. That’s how I started using them myself, and it’s still mostly where I keep them. These days I’ll also use AI to generate some assets. But I keep the parts I enjoy most for myself. That’s modeling, building out scenes, rendering them, then painting over the result by hand.
The process is sacred, not something to optimize away
Specifically in art, I think the process itself is sacred. It shouldn’t answer to the logic of time and money that runs the rest of the market. That’s not a small opinion to hold as someone who works commercially, but I still believe it. To stay in touch with that part of the process, I’ve been making more time for it. Lately, that’s meant more time with traditional painting and drawing, away from the screen. Still, there’s something about working without a screen that keeps the human, unmediated part of making art alive. Also, it’s the part of the work I need to protect from getting scheduled, optimized, or delegated away. I wrote about this at length in a post on flow state in painting. That side of making art feels especially necessary to hold onto today.
Start with friction, and always supervise
The place to start is with the mechanical tasks that generate the most friction. Those are the repetitive, tedious parts of a workflow nobody enjoys anyway. From there, always supervise the output, correct it, adjust it. Treat it as a draft from an assistant, never as a finished decision. That’s the whole point of freeing up mechanical time in the first place. It leaves more of it for thinking and executing with the kind of creativity that’s still specifically human.
AI as a tool for artists, not a replacement for the process
I use AI to build reference material and to learn faster, never to hand over the whole process. Handing over the whole process is where things go wrong, and it usually produces something flat and generic. Nothing about that kind of output gets sharper or more interesting. It just gets faster to produce. That’s the real line for me: AI as a tool for artists who still do the thinking and executing, not a replacement. I don’t know exactly where this goes next, and I stopped pretending I do. But I’d rather learn the tool on my own terms now than be forced to catch up later, from behind.
