Grappling with generative AI as an AI artist
Preface: I aim to keep this thoughtful, and I fully accept that writing about something so divisive is likely to draw criticism, as well as (hopefully) praise. Despite my good faith engagement, some of the naysayers aren't going to like this. And on the opposite side of the same coin, some of you might not like my criticism towards AI art either. With that being said, though, I feel compelled to write this. Not for validation, or followers, but because I genuinely want to process these thoughts and maybe spark some inspiration for others too.
My Journey
I first started generating AI art in, I want to say, April of this year. I started with ChatGPT/DALL-E, and was immediately mystified by what I was able to do with just that. However, human nature got the better of me, and since Sam Altman won't let me generate everything I wanted to (yet), I started looking into local generation and discovered A1111. Going from ChatGPT, which worked right out of the box, to that, provided a huge learning curve. It taught me which models work with what, inpainting, ControlNet, the works. A1111 being what it is, I needed something else. I found ComfyUI and have been a loyal fan ever since. I've moved on from just generating images, to training my own LoRAs now, as well as learning more about RunPod for video generation, as my current local hardware is lacking. I am in LOVE with this artistic process, and I have never found another hobby so creatively fulfilling.
The Problem
All that is to say that I oftentimes found myself at odds with my thoughts on things. One part of me sometimes feels like AI art is just slop and everyone who uses it (myself included) is a talentless hack. and another part says that this is the future and the critics are just losers gatekeeping creativity. But, let's face it, there are some issues. Training data isn't always ethically sourced. Traditional artists sometimes feel ripped off by the AI community for not getting consent before using their work for training. Concept artists, illustrators, and commercial artists are all losing work to AI. Why would a customer pay hundreds for an illustration when you can generate 100 stunning images for pennies? The internet is being flooded "low-effort AI slop", drowning out "real" artists. I won't get into the environmental aspects, but the power needed to train and run these things... It ain't nothing.
Clarity
Here's where things become more clear for me, though. AI models training on scraped images, is arguably comparable to a human studying another's work, taking years to learn someone else's art style, and evolve from there. To me, that comparison is imperfect, but the similarities are there. Pattern recognition, style synthesis. That's what all of this is about, and it's the machine efficiency that scares the other side, and frankly raises new questions we're still grappling with.
Technological displacement has happened in every industry, since forever. The market adjusts, but demand for high-skill, creative work has not and will not disappear, and now traditional artists have begun to adopt AI tools as part of their workflows. The internet was already flooded with low effort content long before the advent of AI. "Bad" art has always existed. The issue is not an AI problem, it's a platform curation problem. Nobody can shut up about AI art, images, videos, etc. because of how high quality it has gotten. The best of the best will always rise to the top. I will say, though, the environment shit... We have to figure that out.
Closing Thoughts
AI art can be a skill, same as traditional art. Deciphering which of the epochs you just trained looks the best, narrowing your dataset, tagging your images, all takes practice. The point I'm trying to make isn't meant to dismiss the criticism. It's real and it's valid. But, I also believe that there's room for mindful, ethical engagement for us exploring AI art. For me that means: creating quality tools for my peers, training my own data, being transparent by never presenting my generated art as traditional art, and not cloning living traditional artist's work. I'm not claiming to have all the answers here. I'm committing to wrestling with these ideas rather than ignoring them.
Yes, the criticisms are real. Yes, there's harm happening. But technology exists now, and we can either engage with it thoughtfully or recklessly. I choose the former. That means acknowledging the problems, operating within ethical boundaries I've set for myself, and being honest about what I'm doing. I'm not claiming to be a traditional artist. I'm learning a new technical skill in an evolving and booming medium. If you are also asking yourself these questions, then I would say that nuance is okay. We can both love this technology while also acknowledging its complications.