The Genesis of Pixel and Mind: The Story of AARON and the First AI-Generated Image
Long before the current era of text prompts, complex neural networks, and instantaneous hyper-realistic renderings, the world of artificial intelligence art began with a mechanical arm, a mainframe computer, and a painter who wanted to understand the very nature of cognition.
While today’s internet is flooded with AI-generated art produced in seconds, the lineage of machine-made imagery traces back more than half a century. The undisputed pioneer of this movement is AARON, an autonomous computer program created in the late 1960s and early 1970s by British-born artist and professor Harold Cohen.
Meet Harold Cohen and AARON
In 1968, Harold Cohen was a successful, internationally acclaimed painter representing Great Britain at prestigious global art showcases like the Venice Biennale. Yet, at the height of his traditional career, he found himself gripped by a radical question: What is the core cognitive mechanism of drawing?
Cohen realized that painting wasn’t merely about physical brushstrokes; it was about a set of rules, spatial relationships, and cognitive decisions. To explore this, he left his traditional canvases behind, moved to the United States, taught himself computer programming, and began developing a software system that would eventually become known as AARON.
Unlike modern AI systems that use deep learning to scrape billions of images from the internet, AARON had no access to a training dataset of human art. Instead, Cohen took a symbolic AI approach. He painstakingly and manually encoded rules of geometry, perspective, human anatomy, and composition into the program.
The Birth of the First AI Image
By 1973, Cohen’s painstaking work culminated in a historic milestone: AARON successfully generated its first autonomous compositions.
The early outputs were primitive by today’s standards—distinctive, elegant black-and-white line drawings. Operating on a connected drawing machine (a flatbed plotter driven by computer code), AARON would autonomously plot abstract shapes, organic patterns, and rudimentary human figures onto paper.
What made these images revolutionary wasn’t just that a computer drew them—computer scientists had previously programmed machines to follow rigid geometric patterns. What made AARON truly artificial intelligence was its autonomy. Cohen did not design a specific picture for the computer to copy; he gave the program a set of rules and cognitive boundaries, allowing AARON to make its own creative decisions about composition, overlap, and scale on the fly. No two drawings were identical, and Cohen himself wouldn’t know what the machine was going to draw until the mechanical pen began moving across the page.
Evolution Over the Decades
As computing power evolved, so did AARON.
- The 1980s and 90s: Cohen updated the software to understand closed forms, allowing AARON to begin filling in shapes with color. Cohen would meticulously paint the physical canvases using color palettes selected by the program, effectively acting as AARON’s human assistant.
- The 2000s: AARON evolved further to generate complex botanical structures, vibrant jungle-like environments, and striking silhouette figures.
AARON’s work transcended the laboratory, eventually being exhibited in legendary institutions such as the Tate Modern in London and the Whitney Museum of American Art in New York.
The Legacy Left Behind
Harold Cohen continued to refine AARON for over four decades until his passing in 2016. He didn’t view AARON as a tool to replace human artists, but rather as an independent entity—an artificial mind sharing a creative partnership with its creator.
When we look at the breathtaking, instantaneous outputs of modern diffusion models today, it is easy to forget the humble, deliberate steps taken decades ago. The jagged black-and-white lines drawn by a mechanical plotter in 1973 proved that creativity was not strictly exclusive to biological organisms. Through AARON, Harold Cohen painted the very first strokes of a new digital horizon.
Leave a Reply