Long before the era of text-to-image prompts, instantaneous generations, and hyper-realistic digital canvases, a British artist and programmer named Harold Cohen sat down in the late 1960s to ask a radical question: What does it mean for a machine to draw?
The answer he forged changed creative history forever, giving birth to AARON—the world’s first artificial intelligence art system—and setting into motion the very first AI-generated images ever created.
The Genesis of AARON: Code as a Brush
To understand the first AI-generated image, you have to strip away everything you know about modern machine learning. There were no neural networks trained on billions of scraped internet photos, no massive GPU clusters, and certainly no prompt bars.
Beginning his work on AARON around 1968 at the University of California, San Diego, Cohen wasn’t trying to build a clever filter or a copying tool. Instead, he attempted to codify human cognitive processes—specifically, the rules of physical drawing, perspective, occlusion, and composition. Cohen translated his own understanding of visual art into symbolic rules. He taught the program how figures relate to one another, how shapes occupy space, and how to balance a composition.
By 1973, following a residency at the Stanford Artificial Intelligence Laboratory, AARON reached a milestone that crowned it a pioneer: it began producing autonomous, original black-and-white line drawings.
What Did the First AI Images Look Like?
If you were to look at the output of early AARON in the early-to-mid 1970s, you wouldn’t see a cinematic portrait or a dreamscape. You would see precise, elegant, and abstract black-and-white line art.
The software ran on primitive computers of the era and drove a custom-built, tethered mechanical drawing machine—a robotic arm holding a pen—that physically moved across a sheet of paper laid out on the floor. AARON made its own real-time compositional choices within the boundaries of Cohen’s rules, meaning no two drawings were completely identical. The machine was literally drawing completely on its own, executing a program that made independent decisions about lines, boundaries, and geometric figures.
As the decades progressed, Cohen continuously updated AARON. By the 1980s and 1990s, the system evolved to handle color selection, painting algorithms, and eventually recognizable organic shapes, plants, and human figures. Work generated by AARON eventually found its way onto the walls of elite institutions like the Tate Modern and the Whitney Museum of American Art.
Why AARON Matters Today
In the shadow of modern diffusion models and Generative Adversarial Networks (GANs)—which synthesize photorealism by crunching colossal datasets—AARON’s rule-based approach might sound quaint. Yet, it represents a purer philosophical breakthrough: the first time a machine was engineered not just to store data, but to execute autonomous creative intent. Cohen didn’t design AARON to draw for him; he designed it to act as an independent entity whose output even surprised its creator.
The journey from AARON’s humble ink lines on a laboratory floor to the vibrant, high-definition synthetic media of the 21st century is vast. But every time an algorithm interprets a stroke, balances a composition, or surprises us with a novel creation, it is standing on the shoulders of a mechanical arm that first started drawing in the 1970s.
Leave a Reply