AI art that's more than a gimmick? Meet AARON

Note: Harold Cohen was a pioneer in computer art, in algorithmic art, and in generative art; but as he told me one afternoon in 2010, he was first and foremost a painter. He was also an engineer whose work defined the first generation of computer-generated art. His system, AARON, is one of the longest-running, continually maintained AI systems in history. Harold Cohen was an exceptional artist, an impressive engineer, and an important bridge between those two worlds.

NEW YORK, NY.- Yes, it’s yet another show of AI-generated art — but wait! The software known as AARON isn’t like other AIs. Its developer, British painter Harold Cohen — being an artist — understood that artificial intelligence isn’t a shortcut to interesting art. It’s a tool, ultimately only as good as its user.

Selections from the paintings Cohen made with AARON, on view at the Whitney Museum, represent their man-machine team’s increasingly sophisticated style. The early pictures, from the 1970s, were limited to abstract, wavering linework and crosshatched blobs — the available computing power couldn’t manage much more — which AARON drew with a robotic plotter and pen. Cohen added patches of blushy, acid color by hand.

Gradually, painstakingly, Cohen deepened AARON’s range to include human figures, objects like tables and flowerpots, and profusions of leafy plants. The 1995 update of AARON could compose jaunty portraits in recognizable, furnished interiors, and color them, using a robot arm to switch between pots of dye. By the mid-2000s, cascades of jagged leaves filled the pictures — in a projection at the Whitney, a version of the software from 2007 builds crayon-hued jungles in real time.

The late 1980s might have been the sweet spot. On view are two examples from Cohen and AARON’s “Bathers Series,” loosely inspired by Paul Cézanne’s impressionist tableaus on the theme. In “Coming to a Lighter Place,” from 1988, the round warbling lines that are AARON’s constant signature inscribe swooshing figures daubed in shades of mustard and powder blue, a spindly forest jolted with tangerine and fuchsia. The painting creaks with fecund joy, as if it wants to go on flourishing.

Before taking up programming, Cohen was an accomplished painter — his canvases, spread with noodly, tractlike shapes, appeared in major exhibitions including the Venice Biennale and Documenta. In 1968, a teaching job at the University of California, San Diego, took him into the fermenting midst of a nascent Silicon Valley and the mushrooming defense industry. The Apple II personal computer was still a decade away when Cohen began tinkering with robotic drawing. He exhibited early experiments in 1972; but AARON proper was born during a residency at Stanford University’s Artificial Intelligence Laboratory from 1973-75. Cohen stayed in California and continued enhancing AARON until his death in 2016.

The current generation of AI image-generating software, from text-to-image programs like Dall-E to splashy animations by Refik Anadol, rely on huge data sets of millions of pictures (many of them copyrighted works by others), which they process and regurgitate. AARON proceeds like a painter: stroke by stroke, following rules for depth and perspective, balanced compositions and color theory, and pulling from a small vocabulary of forms.

AARON has never “seen” a plant, or a human. Instead of imitating a person’s appearance, for example, it constructs figures one line at a time. Its code contains detailed instructions for anatomy, like numbers of limbs, proportions of heads and hands, the locations of joints and plausible postures. At the Whitney, you can see the sketchbooks where Cohen developed this logic, translating movements, like standing up, into code. In one almost mystical schematic, Cohen crisscrossed a drawing of two arms with points and lines like an acupuncturist’s map.

Paintings made with AI-driven robots might sound like a gimmick, especially with the present buzz around chatbots and deepfakes — and the timing of the Whitney show is certainly no accident. But a visit to the galleries dispels that notion, not least because Cohen applied the color in all but one of the paintings; the results are textured and eerily inhuman yet organic — whereas much AI-generated art either lives on a screen or has been flatly printed out. Impressions by Dall-E of work by Jackson Pollock and Lee Krasner appeared last fall at Susan Inglett printed on canvas and awkwardly wrapped around stretcher bars, and fooled no one.

“One of the bargains I made with myself from the very earliest days was that I would never accept the position of having to apologize because this was done by a computer,” Cohen said in a published 1995 conversation with his wife, Becky. “I have always insisted that the work the program did would have to stand on equal terms with art made by hand.”

Today, savvy artists like Seth Price and David Salle are exploring ways to incorporate AI into their practices — to use the software, rather than react to it.

Compared to the visual horrors emerging from the psychedelic meatgrinder of text-to-image AIs like Dall-E, AARON’s docile pictures of people feel friendly and controlled. The Whitney show speaks to a hopeful period of tech development, when the internet’s pioneers envisioned an anarchic realm of the mind, not the boundless attention-gathering machine it became. Cohen developed AARON with intention. The machine and the painter grew together — inefficiently, by tech’s standards, but fruitfully, by art’s. Not to pine for saccharine expressionism or argue for an overly trusting approach to our corporate overlords. But AARON’s purpose-built style of freedom and curiosity seems worth salvaging.

https://artdaily.cc/news/166813/AI-art-that-s-more-than-a-gimmick--Meet-AARON