By Carlo Wolff
Artificial Intelligence is the talk of the media, even the metaverse.
It is far more than the buzzword du jour, and its momentum disturbs, even alarms, some. There are those who deplore it, while others celebrate it. But there is no stopping it, so the choice is between resisting and embracing.
AI is rapidly becoming integral to the creative process, whether you’re a writer, a painter, a mixed-media master or a filmmaker. It scares actors and writers in Hollywood, who went on strike this summer partially in fear of AI’s encroachment. Other kinds of creatives worry AI will take over, essentially sidelining artists of all kinds and destroying such professions as journalism, teaching and painting.
ChatGPT is AI software for writing, based on a large language model type of AI that uses deep learning techniques and huge data sets to comprehend, summarize, generate and predict new content. Part of understanding AI requires learning its vocabulary – and realizing that AI “gives” to the humans that create from it in an asymmetrical exchange. The question is, how is the share determined?
How are local artists responding to AI? What are they learning and using?
The top AI art generation tools are Midjourney, DALL·E and Stable Diffusion, according to AI autodidact Jamal Collins, an East Cleveland-based graphic designer and educator. Midjourney and Stable Diffusion appear to – “appear” because AI is hard to nail down – take text prompts and run them through a diffusion algorithm. It learns, making it fluid, and different from static technology. DALL·E is a neural network, generating new imagery in a process akin to the brain’s.
Collins is the founder of Creative Kids Group, a visual design program for inner-city kids in Akron and Cleveland. He is an AI enthusiast. Not only does he use it in his art, he suggests that refusing to accept it is futile. Nevertheless, if to a lesser degree than the provocative mixed-media artist Kasumi, Collins has reservations.
Image courtesy of Collins.
ONE ARTIST’S VIEW
Here’s how Collins rates current AI technology:
Advantages
• Increased efficiency and productivity in art creation processes
• Exploration of new creative possibilities and styles
• Automation of repetitive tasks, allowing artists to focus on higher-level concepts
• Access to vast amounts of data and inspiration for generating new ideas
Disadvantages
• Potential lack of originality and human touch in
AI-generated art
• Ethical concerns regarding ownership, attribution and plagiarism
• Bias in AI algorithms that may perpetuate societal inequalities
• The risk of overreliance on AI, potentially limiting artistic exploration and personal growth
AI mines both public-domain and proprietary art; it’s all a big pile of data to this voracious software, which doesn’t discriminate among its sources. That bothers both Collins and Kasumi, who lives in Cleveland Heights.
Engaging with AI is one way to keep it honest, suggests Collins, a master of branding and marketing.
A 1997 Bachelor of Fine Arts graduate in design from The University of Akron, he says, “As soon as I stepped off campus, I was outdated,” noting designers with computers left him in the dust. So, he bought a computer and set to learning.
It was the dawn of multimedia. Compact disks, DVDs, MP3s were hot; then came powerful smartphones. Eventually, Collins learned how to edit his own videos so he could build a community “around me and my YouTube channel.”
“The burning question about all of this is ethics,” he says, “and biases and things like that.” Because “it’s taking off so fast, I highly recommend anybody and everybody to get in this, to make sure that it is following the right things it’s supposed to follow, right? So it’s not blurring the lines of copyright infringement and things like that.”
Collins is not so much an expert as an explorer: “I’m still getting in there and learning the pieces and parts, and trying to take away what I could take away and come back and talk about the stuff honestly,” he says.
ANOTHER VIEW
Kasumi has a more wary approach. In her absorbing, kinetic cinema, installations and apps, Kasumi aims to “alchemize” contemporary and evolving technology. A multimedia conceptual artist, she knows how to prospect for imagery for her b-roll fantasias. Ask her where AI ends and the human creator begins, and the answer doesn’t come easily. Kasumi distrusts AI, she explains.
“The line between AI and human art creation can get pretty complex and fuzzy,” she says. “Sometimes AI tools are used as helpers, like assistants not unlike filters and effects built into software used by human artists. They use AI algorithms or software for tasks such as tweaking images, analyzing data, generating ideas or adjusting colors. The AI acts as a creative boost, but the human artist still has the final say and creative vision.”
Another scenario has AI and humans as partners, with the former incorporating the latter’s suggestions into the work of art. “The AI system actively contributes ideas, suggests concepts or generates content, and the human artist incorporates those contributions into the final artwork,” Kasumi says. “It’s a collaborative effort where both the AI and the human artist influence the outcome, blurring the lines between AI and human creativity.”
The obverse might find AI more dominant, shaping guidelines and objectives the human provides “to let the AI generate the artwork on its own. The artist becomes more like a curator or facilitator, shaping and refining the output produced by the AI rather than directly creating it,” she says.
“It’s not always easy to pinpoint exactly where AI ends and the human art creator begins,” Kasumi says. “It can be subjective and depend on the context and the choices made by the human artist throughout the creative process.”
Image courtesy of Kasumi.
THE LEARNING CURVE
Some say AI is the fourth great revolution of the past century, following computers, the internet and the smartphone. What’s difficult to predict is the potential symbiosis between AI and its user, though our bond with the iPhone and its siblings provides an inkling.
“Before computers came into play, before Photoshop and Illustrator, we were doing all of this stuff by hand,” Collins says. “We didn’t even have digital cameras. So to even manipulate the stuff we needed a scanner, right?”
Using Google Images, Collins grabs images of clouds and color patterns. Before Google, Collins had to go to the library for pictures of clouds, or take them himself and then print them himself because the camera wasn’t digital.
“So, you might as well think, jumping from that to being able to just type in clouds and get high-resolution pictures of clouds was almost like AI and kind of revolutionary and changing the aspect of how we created as artists then,” he says.
Such coziness doesn’t sit as well with Kasumi, who suggests in the AI-dominated future, artists could be “prompt masters” with select celebrities ruling the field “with no physical input – a natural extension of the field known as conceptual art.”
Asked for ways in which she incorporates AI into her work, Kasumi says it’s more about “how is AI incorporating my work into it.”
Every image viewable online is fair game to be vacuumed into the gargantuan data sets of LAION-5B used to create “art” by such tools as DALL·E, Midjourney, etc., says Kasumi.
“LAION-5B calls itself a nonprofit organization, but through clever legal means, generates massive profits, much the same way research done at NASA – funded by the tax-paying public – is then used by corporations to generate profit,” she says. “AI does not distinguish between copyrighted and non-copyrighted material.”
Not all AI is doom and gloom for her, however. She now understands mushrooms on what one might call a cellular level.
Kasumi recently experimented with a motion capture suit at Photonic, an animation studio in Mayfield. The owner plugged her into such a suit, five times too large for her, and collected data from her movements. He fed that data into software that used it to generate the movements of a mannequin. In another software, he attached abstract animated objects to those movements.
“Finally, in Stable Diffusion AI software, they came up with the prompts: mushroom, psychedelic and the like,” she says. “The outcome is me moving around as an ever-evolving collection of animated mushrooms.”
