Long before the era of text-to-image generators, billion-parameter neural networks, and instantaneous digital brushstrokes, a quiet revolution began in a laboratory.
While today’s artificial intelligence conjures hyper-realistic portraits and surreal landscapes from a simple sentence typed into a prompt box, the lineage of AI-generated imagery traces back decades to an unexpected pioneer: a British artist named Harold Cohen and a computer program named AARON.
The Birth of AARON: Code as a Canvas
In the late 1960s, Harold Cohen was a successful, internationally acclaimed painter representing Great Britain at prestigious global exhibitions like the Venice Biennale. Yet, at the height of his traditional career, he found himself gripped by a radical question: What makes a drawing a drawing?
Cohen realized that drawing was not merely a physical act, but a cognitive one governed by rules of cognition, space, proportion, and interpretation. Rather than trying to automate a shortcut, Cohen set out to codify human artistic intelligence from scratch. In 1968, he moved to the United States to teach at the University of California, San Diego, and taught himself computer programming.
By the early 1970s—specifically refined during a residency at the Stanford Artificial Intelligence Laboratory in 1973—Cohen birthed AARON, an autonomous computer program designed to create original art.
What Did the “First” AI Image Look Like?
Unlike modern machine learning models that scan billions of copyrighted internet photographs to mimic styles, AARON possessed zero visual training data. It didn’t look at pictures, and it didn’t use neural networks.
Instead, AARON was built on Good Old-Fashioned AI (GOFAI): a massive set of symbolic rules and cognitive procedures written manually by Cohen. Cohen essentially translated his own understanding of visual perception into computer code.
The resulting outputs—first materialized in the early 1970s—were rudimentary, abstract, black-and-white line drawings. The program made autonomous decisions on composition, overlapping shapes, and geometric distribution. A mechanical drawing machine, often referred to as a “turtle” connected to a mainframe computer, translated AARON’s code lines into physical ink on paper.
Watching AARON draw was eerie to early observers. The machine was not reproducing a pre-programmed template; it was calculating compositions on the fly, ensuring that forms balanced one another out dynamically.
Evolution Over Decades
As computing power evolved, so did AARON. Over a career spanning more than four decades:
- The 1980s & 90s: Cohen updated AARON to understand closed forms, allowing the program to handle complex figures, botanical elements, and eventually vibrant color palettes.
- The 2000s: AARON shifted from physical pen-and-paper drawing machines to massive wall projections and screen-based digital environments, generating complex, shifting jungle-like ecosystems autonomously.
Cohen famously viewed AARON not as a tool like a paintbrush, but as a collaborator—an independent entity possessing its own specialized intelligence.
Echoes in the Modern Era
When we prompt a modern diffusion model today, we stand on the shoulders of Harold Cohen’s foundational gamble. AARON proved that creativity—or at least the algorithmic simulation of it—does not strictly require a human hand holding a brush. It requires a system capable of making autonomous choices within a defined creative framework.
While AARON’s primitive 1970s line drawings look quaint next to today’s hyper-polished synthetic media, they represent the true genesis of machine creativity: the exact moment a computer first picked up an imaginary pen and began to draw on its own.
Leave a Reply