23 Oct 2024/53 min/Alejandro Cartagena
MLow X Alejandro Cartagena
A curator with twenty years in traditional art explains why artists took to AI, and how Fellowship's year of daily AI video became a live archive of a movement.
MLow opens by congratulating Alejandro Cartagena on completing a full year of Fellowship's dailies program, which MLow followed closely and collected from. Cartagena, a Dominican artist, curator, educator and publisher based in Mexico, lays out the most developed theory of AI image making in the archive: AI models are trained to describe images to other machines, and inside that machine to machine gibberish a visual language developed that no human could produce, given the scale of data involved. Artists saw a tool never meant for art and, as artists always do, coerced it into doing something it was not supposed to do. He sets AI in a lineage running from photography through video, digital photography and the internet, each one forcing the same question: what is the point of humans in the creative field if a machine can make the image?
The middle of the conversation runs through the dailies program as deliberate real time art history. Rather than wait ten or twenty years for historians to catch up, Fellowship put 137 artists on one shared contract so anyone can segment the year by tool, theme or artist and see how AI video was actually made during its most accelerated year. That leads into a debate MLow raises via his mentor Pindar Van Arman: the difference between AI art and art that merely uses AI tools. Cartagena says most labels will fade, but defends the term for artists like Holly Herndon and Mat Dryhurst, who spent months recording choirs across the UK to train the model that is itself the artwork. From there the two riff on metadata as provenance, set against Cartagena's collection of 50,000 found photographs bought from dumpsters and flea markets, images that lost their origins forever in a way on chain work never will.
The back half turns to the market. MLow connects the AI meme coin mania (the GOAT token hitting a 700 million dollar market cap in a week, Botto at Sotheby's, Turbo) to art's job of showing people what they have not seen; Cartagena admits he will not touch trading but confesses an old hunger, having missed the internet era his computer scientist brothers lived through while he skateboarded. He predicts model building will become the main arena for artistic intentionality, compares it to the hundred decisions behind publishing a photobook, and diagnoses the crypto art market as restructuring rather than dying. His answer is artist development: critical feedback, not pandering, because most artists cannot self criticize, and a shared contract community where every one of the year's roughly 3,200 pieces sold.
What this episode teaches
- AI offers a point of view no human can develop
Cartagena argues AI was trained to describe images to other machines, and that machine to machine conversation produced a visual language built from more data than any human could process. That non human perspective, not efficiency, is what drew artists to the tool.
- Document the movement while it happens
Art history usually takes ten or twenty years to catch up. Fellowship's dailies put a year of AI video experimentation on one blockchain contract as it happened, so anyone can segment it by tool, theme or artist. Cartagena calls that cultural value on top of commercial support.
- AI ends the one style rule, if your voice survives
AI lets an artist develop a style per project and take side roads constantly, which the art market finds hard to swallow. Cartagena says the open question is how far an artist can push that eclecticism and still keep a strong voice through the styles.
- Reserve the AI art label for work where the model is the artwork
Cartagena expects the label to fade into just another tool, as it did with photography. But for artists like Holly Herndon and Mat Dryhurst, who recorded choirs for months to train the model that executes the piece, the model building is the art and the label fits.
- Metadata is provenance photography never had
Cartagena owns 50,000 found photographs with no camera, paper or family information left. Digital work inscribes the model and a timestamp, so MLow's practice of naming the tool and model in metadata gives future found image collectors what analog photography lost forever.
- Intentionality is the scarce thing in the AI age
When anyone can ask a model for a masterpiece, work without intentionality gets discarded. Cartagena compares model building to the decisions behind a photobook: every choice exposes the author, and those choices are where longevity comes from.
- Artists need critics, not cheerleaders
Most artists cannot self criticize, so total independence leads them in circles, repeating their last drop. Cartagena's model pairs artists with people they trust who push back: this repeats what you did five months ago, what if you tried this instead.
- Old infrastructure can be reused for new movements
Web3 discards ideas fast, but Fellowship revived the shared contract, old NFT technology, as a curatorial program. Emerging artists with zero market sat next to established names as one movement, and every piece of the roughly 3,200 minted sold.
The conversation, topic by topic
Who Alejandro Cartagena is
He introduces himself as a Dominican living in Mexico: artist, curator, educator and publisher, with a 20 year career as an artist as his bread and butter. He entered web3 about four years before recording and has spent that time developing other artists and helping them find their voices. MLow notes he collects from Fellowship's dailies, owns Nuclear Samurai pieces, and has a year long collab with Nuclear Samurai in the works.
Why artists took to AI images
Prompted by working with Trevor Paglen, who comes from computer vision research, Cartagena explains that AI is trained to think in images and to tell other machines what things are. In that gibberish a visual language developed, offering a perspective no human can build because of the size of the data involved. He cites AlignDRAW by Elman Mansimov and Gene Kogan's A Book from the Sky as early cases of scientific tools coerced into art.
The photography parallel
Cartagena traces a line from photography to video to digital photography to the internet, with AI combining all of them. Photography triggered abstraction and surrealism as artists went where the camera could not. Now that machines can paint, draw, film and write, he says the deeper question of what humans are for in the creative field is necessary, because art had reached an impasse of pretty pictures and well ordered museums.
AI video's breakout year and MLow's practice
MLow describes a contract he has minted to at least weekly since about November 2023, making videos that mishmash many tools into his own workflow rather than one pure tool. He credits the dailies program as a course guideline that led him to sign up for around 30 AI tool subscriptions. Cartagena says the year absolutely surpassed what anyone thought AI video could do.
Dailies as a segmented survey
The 2023 idea behind dailies was to document the acceleration instead of waiting ten years for history to historicize it. On one contract you can filter for artists experimenting with the city, with Luma, Runway or DALL-E, with narrative, abstraction, gender or family. Cartagena calls it a dream come true for a curator: commercially supportive of artists and culturally valuable for researchers.
Niceaunties and the breadth of the roster
MLow highlights Niceaunties as the one artist who actually shipped work with Sora while everyone else talked about it, plus her Sotheby's partnership. Cartagena says the program deliberately avoided one sub genre of AI video: 137 artists minted into Year One, from a veteran who worked at MTV (a name the transcript garbles) to Niceaunties, an experienced architect whose artistic career is developing live inside the program.
AI art versus art made with AI tools
MLow relays his mentor Pindar Van Arman's distinction: many self described AI artists are not commenting on AI at all, just using it like Photoshop or a camera. Cartagena agrees labels will fade, noting generative AI art is barely nine years old, but defends the term for Holly Herndon and Mat Dryhurst, who spent four or five months recording UK choirs to train their model, and for Diego Trujillo. There the model building is the artwork.
Metadata and the found photograph archive
MLow describes assigning the tool and model in his NFT metadata because the model is more interesting than the AI art label. Cartagena connects that to his eight year collection of 50,000 found photographs bought near dumpsters and flea markets, images with no camera, paper or family information left. On chain metadata means future found image collectors could build archives filtered by model and year, an opportunity analog photography lost.
Reverse engineering images with AI
MLow wonders whether AI could reverse engineer a found photo's origins. Cartagena runs with it: train a model on a fashion museum's dataset of cuts and designers, then send it into the wild to find every Oscar de la Renta from 1981 in ordinary people's pictures. Both agree it would make a strong research project and joke about taking it to the back channels.
AI meme coins and the GOAT moment
MLow notes the AI craze on the meme coin side has not yet rolled into art, citing Botto's Sotheby's result (bids around 70,000, which he reads as high for web3 art), Botto's own token, Turbo, and the GOAT token: an AI shitposting bot trained on Twitter, 4chan and Reddit, built partly on Claude, that hit a roughly 700 million dollar market cap a week after launch. Cartagena says trading is fire he will not touch, but he loves the science fiction of artists training agents.
The old hunger
Cartagena explains why he chases every AI experiment: he missed the development of the internet even though his two computer scientist brothers talked about it constantly while he was off skateboarding. He resolved that would not happen again. AI is his boat for watching things that change the world, even when the experiments live in meme coins and shitposting.
Model building as the next frontier
Asked about trends, Cartagena predicts artists will increasingly build their own models, from scratch or half built. He compares it to publishing a photobook, where paper, type and layout decisions expose the author, and one bad choice can sink good photographs. In the 21st century, when a model can produce a masterpiece on request, intentionality is what separates discarded images from work with longevity.
A market restructuring, not dying
Cartagena cites a poll the collector Pablo Rodriguez-Fraile posted that day asking if the crypto art market is dead or dying; Cartagena answered restructuring. His team's way forward is developing artists rather than just selling art: critical feedback so drop B is not a remake of drop A, because collectors wore out their excitement buying the same thing repeatedly, even from respected artists. Decentralized independence has flaws when artists cannot self criticize.
Lessons from a year of dailies, and the Art Blocks precedent
Cartagena's discoveries: all artists are emotionally unstable, himself included; artists hunger for new things but lack the confidence without trusted people who push rather than pander; and reviving the shared contract let over half the roster, who had never minted NFTs, gain traction through community. All of the roughly 3,200 pieces sold. MLow predicts people will look to Fellowship for AI video the way they look to Art Blocks for generative art, and Cartagena credits Art Blocks as the model they extended with curation and artist development.
In their words
“Here's a perspective and a point of view that no human can develop, because of the intensity of data.”
Alejandro Cartagena“That's like a dream come true for a curator, to work on something that supports the artist commercially but has cultural value.”
Alejandro Cartagena“I also am trying to kind of have a time capsule of now with the tools that I have at my disposal.”
MLow“In the age of AI, intentionality is massive.”
Alejandro Cartagena“All artists are emotionally unstable. That was a discovery.”
Alejandro CartagenaCite this conversation
Alejandro Cartagena in conversation with MLow, "MLow X Alejandro Cartagena," The MLow Show, 23 Oct 2024. https://show.mlow.xyz/e/2024-10-23-mlow-x-alejandro-cartagena.html. Quoted from a machine transcript.
Quotations come from machine transcripts. Check wording against the audio before republishing.