AI Art as Curation

AI art as curation

Asking whether AI art involves real creativity mostly misses what's interesting about it

Andy Masley

Jul 03, 2026

When I make an AI image or song, I don’t feel the kind of satisfaction I do when I draw, or play something on the guitar. Angry memes about how “no one making AI art is an artist” just bounce off me, I never felt that way in the first place.

But AI images and music have turned out to be pretty emotionally resonant with me at a level I didn’t expect, for another reason. They let me curate very specific vibes that matter to me and would be hard to assemble anywhere else. Curation isn't creation, but it's a separate and (I think) equally valuable outlet. A way to get at intuitions that are hard to put into words but feel important to communicate.

My basic experience of creating an AI image is like finding a neat rock. I’m not responsible for the shape of the rock at all, but I might be interested in putting it into an arrangement with other rocks and showing people my specific taste in neat rocks. Maybe I get some points for knowing where to look for the rock, for knowing broadly what types of rocks appear where. Playing with an AI model is like searching its high dimensional vector space for neat rocks, with a taste for specific rocks I can only find in specific places.

It’s also like making a music playlist. When I put a playlist together, I obviously know I didn’t create the songs themselves, but I’m hoping that by associating with them, and combining and contrasting them in specific ways, I can get across some general sense about the world I think is important. Typing a very specific style prompt into Suno and then playing around with the result feels pretty similar to creating a playlist. I’m venturing out into the vector space of the model to try to find extremely specific sounds I value a lot to add to it. I’ve had way more success here than I expected. The first 2 minutes of this collection of shoegaze songs Suno made for me is a very specific sound I value a lot and want to associate with, especially the big explosion at 50 seconds. It gets across a very specific inarticulable sense of openness to the world I had when I was a teenager. I feel like if I could just glue this sense together with specific images and smells and memories and vibes, people would really “get” me.

No AI music has reached the same level of emotional resonance for me as my favorite bands. I like the sound of the shoegaze songs I made a lot, but the best human-made albums in the genre are in another stratosphere, and I don’t expect AI to approach this quality soon. Suno’s already showing too many signs of being too optimized for more general mid taste. Still, comparative advantage exists. The best human artists remain in a completely separate league from basically all AI music for me, but I don’t always want to listen to the best stuff. Sometimes I want a very specific sound I haven’t heard before in a genre I like, and the best shoegaze album was so good it kind of singlehandedly destroyed the genre, so there’s less human made shoegaze to pick from than I’d like. AI gives me a way of reaching out into the vector space of all possible mid shoegaze to say “This… this is what I’m into, this ‘gets’ me” and the experience of compiling said mid shoegaze in one place is rewarding in the same way making a playlist is rewarding. For even less explored genres I love like dungeon synth, AI can fill the gaps I feel as I explore the very limited but great set of what humans have actually produced in the genre. None of this feels like I’m creating the music, but it does feel satisfying to put these things I’ve found together in specific ways, to know where to look to find things I like, and to use my past great experiences with music to try to express my taste with the new sounds I find.

Last year I posted a big collection of AI images I’d made that I liked. Again, I didn’t feel artistic ownership over these images. I didn’t feel like my creativity had gone into them. And yet I did feel like I’d been creative in how I used them to express my taste, in the same way making a playlist can involve creativity, and the post resonated, where multiple readers reached out to talk about how our tastes were way more similar than they had expected. This was a fun experience of feeling a bit like discovering far off people secretly in the same tribe, who were moved in the same subtle ways by similar experiences. It was fun to find this together, and the way I found it was purely via what I consider curation, not creation, and that was really valuable on its own!

I really love finding ways other people have curated what they like and exploring them. The sense that other far away people have resonated with specific art or vibes or other media makes the world feel like it has incredible depth. The world can often feel lonely and flat. Maybe the most flat-feeling image I’ve seen is the famous Breezewood PA picture:

The polar opposite of this image for me is a good curated list of things specific people like, either on their personal sites or Goodreads or Letterboxd or music app. These always feel like the world has so much depth, so many points of surprising contact between what we experience alone. I’d made a long list on my personal site of media I like mainly to just imitate other bloggers who made dense explorable curated lists of media they love. I could spend hours poking through these.

The greatest curator on the internet is Piero Scaruffi, who over 30 years has reviewed basically all music on his website that looks basically identical to when he started it in the 90s. It’s my favorite example of curation elevated to a monumental complex achievement:

Scaruffi’s a cognitive scientist who’s focused on AI a lot and among other things also has what I think is the funniest subtitle of any book on AI:

I mutter “Life in the coming age of incredibly stupid machines” once a week, though more and more as cope.

Scaruffi seems to have reviewed most other culture as well, and also publishes long dense essays on things like the history of knowledge, and curates other lists like his favorite waterfalls.

I’d recommend poking around his website for a while, maybe starting with his reviews of his favorite albums.

Scaruffi has a short (and self serving) post on the “Meaning of Art”:

This flips the status of creation and curation: most creation is really just imitation, raw material for curators to build statements about the world.

This seems a little extreme, but consider how many albums you listen to at least in part to be the type of person who listens to them. How often when you listen to music do you feel somewhat ascended from your everyday life because you’re part of the secret community of people who listen to it, the people who know the canon of truly great music? How often do you feel like listening to specific new music will ensure you’re one of the elect?

The curation of rival canons of great music really exploded as the internet took off, and canons reacting to each other to distinguish themselves, and to rise above all others, got faster and faster. For example, if you were on the 4chan music board /mu/ in the early 2010s, you probably liked a lot of these albums:

I know I did! I worry that when I talked about shoegaze earlier, I added a link to Loveless partly to say “Of course I’m allowed to listen to AI shoegaze, because I know about and am deferential to the truly great shoegaze canon, I’m not some pleb.”

But the existence of these canons encouraged rival canons to pop up, where you could not only feel more “in the know” and ascended compared to most people, but even more in the know by people who thought they were in the know. Take this other list someone on 4chan made, and especially the note at the bottom.

Yerself is Steam is a great album, but in my opinion it’s clearly not nearly as good as Loveless, which the author probably considers /mu/core and “radio music.” This is basically a hostile act of curation, using the art in the way Scaruffi describes to say something about the values and ideology of the critic, separating the critic from rival canons. This becomes more and more far removed from the art itself and is instead part of a drawn out contest of curation.

Once you start paying attention to what curators are trying to do, and how often any specific art is more of a stepping stone to say “I am with the cool good authentic community of people, by listening to my curated canon you will ascend to them too” it’s hard to look away. So many Pitchfork reviews were much more about the type of people listening than the music itself. Look at their 10/10 review of Kid A, how much of it is referencing other key parts of the canon, or the interesting location of the reviewer. Or their 0/10 review of Bachman Turner Overdrive’s greatest hits, which very specifically says “If you like this music, you are not one of the elect”:

Those who still cling to BTO-- and you know who you are-- are usually dorky minivan types who think rock and roll is about having your friends over for beer and barbeque. Let me say on behalf of the world's young people that all you people truly suck. Even in your retro phase there's nothing remotely charming or rocking about you. Here's hoping that next T-Bone you eat leaves you choking on your own excess. Die! Die! Die!

Seems a little much…

This is all to say that curation on its own can be a monumental and impressive task, that it’s incredibly fraught and bound up with our self conception and rival systems of status, but also at its best can be a unique satisfying bridge between people, all separate from any act of creation of the art itself. The idea that AI radically expands the space of curation and opens up ways for individual people to be much more specific with the images and sounds they collect and display to the world as “what they’re about” is under discussed in the conversation about AI and art. Most AI art conversations I see online are still reduced to whether AI art “involves creativity.” For me it involves as much creativity as assembling a playlist. Whether that’s like artistic creativity seems less interesting to me than what it means about the future of how people share their taste with each other, and use it to either find or compete with each other. How much of the good feelings we get when we find others who like the same art is from authentic shared experience vs just the camaraderie of identifying other members of our ascended rival tribes is a pretty foundational question about human life I don’t have a clear answer to. I worry that a lot of people in the AI and art debate fail to appreciate how interesting curation is as a separate concept, and the heights of achievement and depths of emotional turmoil it can reach.

I’ll close with a few AI images I made recently:

Some Massachusetts triple decker art:

Nostalgic photographs:

Geodesic domes: