Before I get to my Weekly Wrap-Up, I need to address a question of burning relevance to all Substackers today. It’s time to talk about… AI slop. More specifically, AI-slop art.
It’s obvious that I use an LLM (Gemini) to create most of the art for this Substack. I’m open about it, but I do know that many people have strongly negative feelings about that. Here’s what happened in a nutshell.
In the first week of Christendom Reborn I was having a lot of fun playing with AI art. Then I got cold feet. The outcry against “AI slop” was deafening, making me feel I might be creating the world’s cringiest Substack. So I shifted gears and spent a few weeks finding art on Wikimedia and other online sources, which is old hat for me as an editor at Law & Liberty. I do know how to find art online. But even though I was pleased with some of the things I found, on the whole it felt boring and uninspired. When you’re illustrating someone else’s work, you have to play it somewhat safe, but I thought perhaps a personal Substack could afford to be a little more creative, more daring, with more of a distinctive visual brand. I can use the AI to capture some of the images flitting through my own mind while I’m writing. Each week as I choose my theme, I think about what kind of artistic strategy would set the right mood. It’s satisfying to use visuals that way. Perhaps a little bit interesting to readers, too?
At the same time, I started to get irritated by all the people who complained about “AI slop” and then when I went to their sites… they just seemed to have a lot of super-random photos of houseplants or cats or restaurant meals. All The Random is fine, as long as it’s not AI slop? Seems silly.
To be clear, I’m only focused here on AI-produced art. All the written content of Christendom Reborn comes directly from my own brain and fingers, though I do use LLMs for research, and when I’ve finished something I always drop it into Gemini with the instruction, “Fact check and critique.” I’m still mulling over my views on AI-produced prose, but the critical point is that I am a writer, looking to communicate my ideas to other people in written form. So that’s what you get here. A human talking to humans.
Art is different. I am not a visual artist. I don’t expect anyone to follow Christendom Reborn for the art. However insofar as online sites have a visual component, it seems to me that the best I can do is to capture as nearly as possible what’s in my head as I’m writing, which may have some relevant symbolic significance, or humor value, or just a mood that might accent the reader’s experience. I wouldn’t be able to draw it. But Gemini can.
It’s interesting sometimes having a back-and-forth with my LLM over those images. Sometimes we fight a bit! When I wrote “Sex is Not Safe,” for instance, Gemini tried to get me to use a sweet little image of a bridal couple under an arch with hearts and flowers everywhere. I thought that was hilariously contrary to the central point of the piece, so we duked it out a little and ended up with… something different. I liked that one.
I designed this Substack in about three hours on a Sunday evening, together with my kids. It involved some heartwarming hilarity, as for instance when Gemini randomly threw a dragon into the welcome page image, and my 11-year-old argued vociferously for keeping it. But all things considered, I was pretty happy with it, and I still love my green and white chi rho logo. It seems to me like a real blessing that AI tools that enable solo writers to build a solid brand for ourselves, in a way that used to be possible mainly for entities (e.g. corporations) with the resources to hire graphic designers.
So I’m owning it. This is a Substack with AI-slop art. It’s fun, I like it, and it seems to me like the clear best option all things considered. We’ll see how things go moving forward.
I’d be interested in other people’s views, however! I’ll set the comment feature on this post to “Everyone,” so that you can leave your thoughts even if you’re not willing to subscribe to an AI-slop-driven Substack.
On that note, here was the week:
Sunday I transitioned from Traditionalist Week with a brief reflection on prayer.
On Monday, I set up this two-week series on “Why Christian Philosophy” and said a few words on ancient non-Christian faiths and how I think about them.
Tuesday I posted my essay on “What is Philosophy For?” exploring the question of what one should look for in a philosophical view. That was followed on Wednesday with some recommendations for other philosophy Substacks.
My essay on “Two Workhorses of Christian Philosophy” posted Thursday, explaining why hylomorphism and natural law are, for the Christian tradition, like the “simple machines” of classical physics. That led to Friday’s re-up of a National Review piece on “A Philosophical Supervillain” and how he was saved from oblivion by a pair of Italian philologists. Amazing story.
Many thanks to Nathan Smith, Dan Hugger, and Reepicheep for their commentary on the site! I have Reepicheep to thank for this fun supplement to a philosophy series. Greeks win! Greeks win!
More philosophy next week. Time to talk about modernity.
Okay. So, I don’t personally use AI art, and on my substack posts I usually try to find a decent painting from the public domain or a photograph that I took of some beautiful nature scene.
However, after discussing what kind of art AI art might be over breakfast with Mark, we came up with an idea that making images with AI is kind of like making a collage. AI has images that have been fed to it and with your imagination and prompting puts the images together in the way that you desire. Kind of like cutting up magazines and gluing images together into a new one. When I think of it that way, it feels palatable to me.
Thanks for setting that out. I agree — AI art is very often appropriately illustrative of ideas that one is trying to set out clearly for other people to understand. And, like you say, sometimes it is way off the bat at first, but you can tweak it until it fits. That’s truly liberating, especially for those of us whose primary skill lies in verbal communication. And AI is a good editorial critic too, although its sentence construction is a bit clunky. I never engage with texts that are clearly AI-generated, since if I wanted that, I can write perfectly good prompts of my own.