The Pixel Genesis: How Harold Cohen and AARON Created the First AI-Generated Art

If you type a prompt into an image generator today, you can conjure photorealistic landscapes, oil paintings, or surrealist dreamscapes in mere seconds. It feels like magic, a modern lightning strike of technology and culture. Yet, the story of artificial intelligence making art does not begin in Silicon Valley tech labs during the deep learning boom of the 2010s. It traces back more than half a century, born in the quiet hum of mainframe computers in the late 1960s and early 1970s.

Its creator was not a computer scientist looking to disrupt an industry, but a celebrated British painter named Harold Cohen.

The Painter Who Turned to Code

In the late 1960s, Harold Cohen was at the top of his traditional art game. He had represented Great Britain at the prestigious Venice Biennale and exhibited globally. But Cohen developed an existential curiosity: What is an image? What fundamental rules govern how a human hand translates thought into a visual composition?

To answer this, Cohen abandoned his brushes for a time and relocated to the University of California, San Diego, to study computer science. He wanted to see if a machine could learn to draw independently.

By 1973, Cohen had birthed a software program that would change art history forever: AARON.

Meet AARON: The World’s First AI Artist

AARON was unlike any digital tool that came before it. Early computer graphics pioneers like Georg Nees and Frieder Nake were writing code to generate precise geometric patterns. But AARON was different. It was designed to possess autonomy—to make its own creative decisions.

Operating on primitive mainframes, AARON did not use neural networks, billions of scraped parameters, or diffusion models. Instead, Cohen hand-coded a sophisticated cognitive rulebook into the software. He taught AARON the rules of human anatomy, perspective, object overlap, and spatial balance.

When AARON ran, it did not follow a rigid, pre-drawn template. Within the boundaries of the rules Cohen provided, the program decided autonomously what lines to draw, how to compose the figures, and where elements should sit on the page. Connected to a mechanical drawing arm—a flatbed plotter holding a pen—the computer would physically move across paper, slowly scratching out surreal, organic shapes and human-like figures all by itself.

The resulting black-and-white line drawings marked a monumental milestone: the first time a machine had generated original visual art through independent decision-making.

Evolution of a Silicon Mind

Over the next several decades, Cohen didn’t abandon AARON; he watched it grow up. As computing power expanded, so did AARON’s capabilities.

In the 1980s and 1990s, Cohen updated the software to understand color. AARON stopped relying solely on human operators to paint its lines and began filling in its own complex color palettes, guided by algorithms that understood how colors interacted visually. By the 2000s, later iterations like Aaron KCAT were projecting sprawling, jungle-like environments onto massive gallery walls, mutating and creating unique botanical ecosystems in real time.

For decades, Cohen and AARON existed in a fascinating philosophical space: was AARON the artist, or was Cohen the artist using a very sophisticated brush? Cohen himself often argued that AARON possessed a form of independent cognition, operating with a creative agency that surprised even its maker.

The Long Shadow of AARON

Today, we live in an era dominated by Generative Adversarial Networks (GANs), Transformers, and massive text-to-image models. Millions of people daily command machines to paint in the styles of historical masters or invent entirely new aesthetics.

Yet, the core philosophical debate—and the thrill of watching a non-human entity manifest visual imagination—remains identical to what Harold Cohen witnessed in his studio in 1973. Long before GPUs rendered hyper-realistic portraits, a painter and a mainframe proved that art was not strictly bound to human biology. They proved that creativity could be codified, calculated, and brought to life through code.