An Artist Invited Blind People to Use an A.I. Image Generator. The Unsettling Results Could Help Make Art More Accessible

Researchers tested the limitations of images created with text prompts.

Auguste Rodin, The Three Shades, before 1886. Bronze. Garden of Sculptures. Rodin Museum, Paris, France. Photo by: PHAS/Universal Images Group via Getty Images.

A project organized by artist Cosmo Wenman experimented with the use of artificial intelligence for blind users and has raised questions about whether the advancement in generative A.I. technologies can aid in accessibility for artists who are blind.

Wenman, known for his 3D design and fabrication art, worked on the quasi-academic art project with blind researchers Brandon Biggs, Lindsay Yazzolino, and Joshua Miele. The team started their work last fall and generated 4,110 images from text prompts input to Midjourney.

In describing the origin of the project, Wenman said he spent the last decade creating tactile exhibits and working with companies specializing in universal access, often consulting with researchers who are blind themselves to provide feedback for that design cycle.

“For the last year when these A.I. solutions started becoming available, I was just naturally interested in how this might interact with accessibility issues,” Wenman told Artnet News.

He said the ability of A.I. models and their creators to mine human art and imagery has created a “new form of accessibility.”

“With photography or even 3D scans of sculptures, we access a surrogate of that thing. With A.I., we are accessing elements of the thing broken down for us and more readily available for remixing,” Wenman said. “It’s astonishing. If a museum published a photograph, you could download it. But you’re not downloading understanding of it or breaking it down into concepts. A.I. does that.”

Wenman said programs like Midjourney find “weird and thematic intersections” in art that are not obvious to human eyes. He pointed to pictures generated from prompts written by Miele that share visual language with the sculptures of Auguste Rodin. Miele had attempted to design his own works in Rodin’s style based on his knowledge from having touched a few of the sculptor’s works.

1) A grotesque illustration of a baby’s head next to a variety of amorphous objects that resemble severed arms and legs, some with fused hands or fingers. They are surrounded by a mass of blood red veins and tendrils. 2) A photograph of 30 to 40 plaster limbs, which Auguste Rodin referred to as “giblets,” lying together in a messy collection. There are gesturing hands, forearms outstretched or bent at the elbow, and a few bones resembling human femurs. Some of the arms appear to be life-size, and many are much smaller. They are all loosely modeled, with the smaller arms being strangely distorted and lumpy.

Left: Joshua Miele via Midjourney. Right: Auguste Rodin, Spare Arms, Hands, and Bones. Images courtesy of Musée Rodin and Cosmo Wenman

Rodin: a small, unfinished sculpture of a baby, roughly modeled in low detail with irregular surfaces, dull brown and green. One arm extends out sideways side, missing its hand. The other arm crosses its chest, without its forearm. It has a rough, patchy, unpleasant appearance. Its face looks directly out to the viewer. A second image shows its ragged unfinished torso. Miele: An unpleasant image of a fetus or newborn baby with an angry scowling expression and glassy yellow eyes staring at the viewer. Its arm is mishapen, its wrinkled hand grasps a larger fleshy feature that may be an arm. Rodin: an unfinished model of an infant in grainy terracotta, its face very roughly modeled. Its eyes and ears are merely impressions on the surface of its face, and its nose and mouth are jagged and very irregular. Its left arm is outstretched, and severed in the middle of its bicep. Rodin: a rough, miniature plaster cast of a lumpy arm, bent at the elbow. The hand is open and its fingers are modeled in very low detail

Second from left: Joshua Miele via Midjourney. Left and top center: Auguste Rodin, Child with barely modeled body. Bottom center and right: Auguste Rodin, Angled Left Arm, and Child sitting on a mound. Images courtesy of Cosmo Wenman, Musée Rodin and Jérome Manoukian

1) Miele: a naked body crouches on a tree branch, its legs distorted, bent in the wrong directions. One arm extends down to grasp a branch. Its legs may morph into hands that also grasp at the branches. All the hands, or possibly feet, are highly irregular and have too many fingers. At its shoulders, its head is difficult to discern but appears to hang down, unnaturally low, creating a horizontal upper edge that crowds the top of the image. 2) Rodin: a tan-colored plaster sculpture of three identical nude male figures arrayed side by side. They stand with their necks bent down at unnatural angles and their heads hanging low, below their shoulders. They are positioned in a semi-circle, so their heads meet, huddling together. Each figure’s left arm extends down to meet the others’. Their other arms hang at their sides, broken at the forearms; their hands missing. The way their heads hang at the shoulder creates a horizontal sightline that makes for an unusual, crowded, strained gesture.

Left: Joshua Miele via Midjourney. Right: Auguste Rodin, The Three Shades (1886). Images courtesy of Cosmo Wenman and Musée Rodin.

“Thick limbs and stubby fingers, short men and twisted, interlocking partially embedded baby body parts, straining and struggling despite inevitable frustration and failure with infernal forces held back against barely better odds,” Miele wrote in his prompt.

The pictures output by Midjourney looked grotesque and gory in ways that do not obviously evoke Rodin’s best-known works like The Thinker (1904) and The Kiss (1882).

“I didn’t think Josh meant to create this visceral bloody imagery, and if he was using it on his own, he wouldn’t have known he had done that,” Wenman told Artnet News.

He shared images of Miele’s pieces, which seem closely inspired by Rodin’s oeuvre when examined more thoroughly.

“Crowded, fused, and misshapen embryonic nodes shadow the distorted silhouettes of menacing, hyperextended hands with painfully cramped fingers of plaster,” Wenman wrote in a blog post about the project. “Rodin’s exploration of the expressivity of irregular, oddly posed, and loosely modeled hands seems simpatico with Midjourney’s much-derided emphasis on gesture over accuracy, the emphasis Miele’s text clearly called for.”

The first image presents an impressionistic painting of a crowd of people in blue and white gowns surrounding a taller figure wearing a red cloak. The faces are not clearly visible, and the scene is set at night with vague trees and a partially obscured moon. The second image features a stylized portrait of a man with black hair and a beard, depicted in low detail. He wears a multi-colored robe, and another figure is abstractly represented behind him. In the third image, a loose, gestural portrait shows several faces arranged in a circle around a central man with a beard. Bright candles and objects interspersed between them create a medieval iconography-inspired composition. The fourth image depicts a detailed nighttime scene with a crowd gathered around a robed man. He has a halo-like light behind his head, evoking a sense of divinity, and the painting has a semi-realistic and narrative quality commonly found in contemporary Christian-themed art.

Lindsay Yazzolino. Four images generated by Midjourney seeking to replicate Leonardo da Vinci’s The Last Supper. (2023) Images courtesy of Cosmo Wenman

Detail of Jesus’s disciples facing outward toward the viewer. Along the bottom of the image are the darkened backs of people’s heads, who are themselves viewing the painting with us.

Lindsay Yazzolino. A detailed look at one of four images generated by Midjourney seeking to replicate Leonardo da Vinci’s The Last Supper. (2023) Images courtesy of Cosmo Wenman

One grouping of images came from a prompt written by Yazzolino, who described Leonardo da Vinci’s The Last Supper (1948) in an attempt to have Midjourney replicate it. Her prompt reads, “famous painting depicting Jesus with his followers during the night before he is put on the cross.”

“Even with the Jesus imagery, they don’t look like Leonardo, but are evoking themes that can be stirring if that’s what you’re looking for,” Wenman said.

As the only seeing person on the team, he said he was “fascinated” by the results — many of which output visual illusions the artist had not anticipated.

“The Byzantine icon imagery of Jesus with his apostles, where they look like the silhouettes of the audience in front of them, I did not expect that,” Wenman said of one image output by Yazzolino’s prompt. “But looking back at what Lindsay described, it includes the idea of a ‘famous painting’ which implies people looking at it.”

Michelangelo’s David and four images of white stone sculptures of partially clothed, muscular figures standing in dramatic poses. 1) David, white marble sculpture of a nude man with a fit, muscular physique, standing upright, looking to his left. His left hand is held up near his left shoulder, holding a sling. His right hand hangs at his side grasping a small stone. His posture and expression convey calm readiness and determination. 2) and 3) Abstracted, roughly textured stone. One figure has its left arm raised with his hand behind his head, with a satchel hanging at his waist. The other looks to his left, with a strap diagonally across his torso. 4) Detailed physique of a modern body builder, looking to one side. A length of fabric diagonally across his chest wraps around and supports his left forearm. 5) Detailed, muscular figure. The right arm is bent up so his hand touches his right shoulder, grasping another piece of fabric. His head is turned to the left. His eyes look to the distance.

Left: Michelangelo’s David (1504). Center and right: Brandon Biggs via Midjourney. Photo courtesy of Cosmo Wenman

1) The Mona Lisa. A woman sits upright with her arms folded, looking toward the viewer with a very subtle, cryptic smile. The landscape behind features a road, a bridge, and further on, green hills and calm water that transition to faint, icy features. The painting is rendered softly with unusual lifelike effect. 2) and 3) Paintings of women from the chest up, with neutral expressions. Streaks of reflection run across the paintings as though they are behind glass. 4) Nearly photoreal close-up image of a woman’s face. She looks at the viewer or slightly to one side with a worried expression. There are watery reflections across her face, as though she is seen through distorted or wet glass. 5) A close-up image of a woman’s face from the neck up to her forehead, nearly photoreal. Her head is very slightly turned to the side, and she is looking either at the viewer or slightly to one side. She has a neutral expression. There are reflections across her face, as though she was photographed through glass.

Left: Leonardo’s Mona Lisa (1519). Center and right: Brandon Biggs via Midjourney. Photo courtesy of Cosmo Wenman

A.I. companies like Midjourney are openly talking about the issues, wanting to have precision and trying to give users more control over generated results based on the composition of their text input. Such advances are certain to make these tools more accessible for blind users, but could take away some serendipitous beauty. Wenman said chaos leads to visual intrigue when artists who are blind use image generators to create art.

“I hope that all these systems retain the ability to have some open-ended chaotic aspect to it,” Wenman said. “In Midjourney there is a command called chaos, a chaos factor of 0 to 100. That’s definitely really compelling but less utilitarian if you want a specific image.”

Wenman said A.I. companies will face challenges in closing an open loop that occurs where a blind user cannot vouch for the fidelity of the end result.

“The image technology can get better but a blind user is still not going to be able to evaluate the image,” Wenman said. “Josh talked about agency and how technologies enhance agency, as opposed to asking someone for help. It’s a touchy subject in blind issues, so the more a blind person can rely on something like text-to-image to be precise, the more they might find some value in expressing themselves.”

Eight images of unusual animals. One shows a small dog-like creature with large ears and eyes, black fur with white spots, and an extra hind leg. Another depicts a creature with a dog-like body, four long, thin legs, and an elongated nose or horn. There's a small creature with a dog-like body, short wide ears, and a single horn on its forehead. One shows an anatomically unique animal resembling a combination of a dog, gazelle, and giraffe, with a chihuahua-like head, a short horn, and black polka dots. Another shows a peculiar creature with three legs, a feather-like tail, and a long neck tapering to a point. There is a horse or dog-like creature with three legs, a giraffe-shaped head, and a spotted wrinkled fabric or fur. Another shows an implausible animal with two long legs in front, a thick leg in the back, and a horn-like protrusion above its shoulder. Lastly, there's a depiction of a giraffe with minor imperfections, such as an extra horn-like protrusion and an additional tail.

Top row: Joshua Miele. Bottom row: Brandon Biggs. Via Midjourney. Eight images of giraffe-like creatures output my text prompts written by the two researchers. Photo courtesy of Cosmo Wenman

Eight images of creatures with elephant-like features. One presents a wooden or leathery creature standing on tree-trunk appendages, featuring a curved tusk and a trunk-like appendage. Another depicts a creature with elongated legs, curved tusks, and no ears. One is a mix of elephantine traits, including wing-like ears and trunk-like appendages, resting on a single leg. Another shows a creature with odd-shaped limbs, a pointed head, and a long trunk-like extension. A taxidermy-like creature resembling a decaying elephant with a curved tusk is shown on three legs. There is a creature with a moose or rhinoceros-shaped head, an upright horn, and some elephant-like features. Another shows a wildebeest-like animal with ram-like horns and distinctive lips, accompanied by a floating curving horn in a savanna environment. Lastly, an African elephant with very slight proportion deviations, like a thin trunk end, extra short tusk, and additional toes, is seen in a savannah-like setting.

Brandon Biggs. Eight images of elephantine creatures output by Midjourney from the text prompt “a large 4-legged animal with a long trunk, large ears, long tusks, hairless hide, and heavy.” Photo courtesy of Cosmo Wenman

Companies like Midjourney are trying to close that feedback loop by creating the ability to generate text from an image, which can already be done with models like OpenAI’s GPT-4. Such a back-and-forth system will get better and provide blind users with some feedback on generated images.

“When everyone can make beautiful imagery using these tools, who’s looking at it and why we’re looking at it is going to become more relevant,” Wenman said.

He said the team plans to curate their images into an exhibition and has discussed collaborating again to experiment with text-to-3D software. The artist plans to fabricate in bronze the 3D designs generated by the team’s text prompts.

“I think it’s important technology,” Wenman said of generative A.I. “I think the larger scale accessibility question, being able to integrate the thematic elemental make-up of a work, is really extraordinary. It’s essentially moveable type for visual rhetoric. That has never happened before.”

More Trending Stories:  

An Elderly Couple Sold a ‘Worthless’ African Mask for $157. Now They Are Suing the Buyer Who Auctioned It for $4.4 Million 

A Norwegian Man Stumbled Upon a Trove of Gold Dating to the Early Middle Ages, Including a Rare Pendant Depicting the Norse God Odin 

A Top Antiquities Sleuth Has Called Out the Manhattan D.A. For Continually Passing His Work Off As Its Own 

Emerging Artist Li Hei Di Calls Her London Studio a ‘Parallel Universe,’ Where Hong Kong’s Cinematic Heroines and Mystical Abstraction Meet 

After Its Team-Up With Pokémon, Scalpers Swarm the Van Gogh Museum to Snap Up Merch and ‘Pick the Gift Shop Clean’ 

An Enigmatic Still-Life Picasso, Made During His Now-Celebrated ‘Wonder Year’ of 1932, Will Hit the Auction Block This Fall 

Get a Closer Look at Lagos-Based Artist Nengi Omuku’s Intricate Textile Paintings—Made on Traditional Nigerian Cloth 

A Mexican Journalist Went Viral After He Presented ‘Alien Bodies’ to Congress. Now He Is Accused of Plundering Them From Ancient Sites 

Follow Artnet News on Facebook:

Want to stay ahead of the art world? Subscribe to our newsletter to get the breaking news, eye-opening interviews, and incisive critical takes that drive the conversation forward.
Article topics