Before the Prompt: The Untold Story of AARON and the First AI-Generated Images of 1973
Long before modern text-to-image generators could render hyper-realistic photographs from a simple sentence, the birth of artificial intelligence art took place in a quiet university laboratory. The year was 1973, and the tool was not a deep neural network humming on cloud GPUs, but a revolutionary piece of software called AARON.
Conceived not by a computer scientist, but by a distinguished British painter, AARON’s creation permanently blurred the line between human creativity and machine logic.
The Artist Who Invited a Machine Into the Studio
In the late 1960s, Harold Cohen was a celebrated painter representing Great Britain at elite global exhibitions like the Venice Biennale. Yet, at the height of his traditional career, Cohen found himself consumed by a profound philosophical question: What are the minimum conditions under which a set of marks on a surface functions as an image?
To answer this, Cohen made a radical pivot. He left behind the brushes and canvases of the traditional art world, moved to the United States (eventually becoming a professor at the University of California, San Diego), and taught himself computer science. He wanted to strip image-making down to its cognitive bones.
By 1973, those years of programming culminated in a breakthrough: AARON, widely recognized as the world’s first autonomous art-generating software system.
How 1973 AI Art Actually Worked
To modern eyes accustomed to scraping billions of internet images through diffusion models, AARON’s architecture feels alien. It did not “learn” from a dataset because digital photography and the internet did not yet exist. Instead, AARON was a symbolic, rules-based AI.
Cohen painstakingly translated his own understanding of visual cognition into raw code:
- Structural Rules: AARON was given logical boundaries regarding how physical objects occupy space, how lines intersect, and how shapes overlap.
- Autonomous Decision-Making: Once Cohen initiated the program, he did not guide the mechanical hand. AARON decided composition, scale, and placement entirely on its own, utilizing abstract lines, closed forms, and primitive geometric elements.
- The Drawing Machine: The software outputted instructions to a custom-built, turtle-like drawing machine connected to the computer, which physically held a pen and moved across a flatbed paper surface to execute the design.
The resulting 1973 images were striking minimalist compositions—abstract, intricate line drawings that possessed an organic, almost tribal rhythm. They were unmistakably artistic, yet entirely non-human in origin.
A Lifelong Collaboration
What makes the story of AARON unique is that 1973 was only the beginning. Cohen did not treat AARON as a one-off experiment; he spent the next four decades expanding its capabilities.
As computing power evolved, Cohen updated AARON:
- In the 1980s and 1990s, AARON evolved from drawing abstract black-and-white lines to generating complex, recognizable organic forms like rocks, plants, and human figures.
- Eventually, Cohen taught the program how to handle color, carefully writing algorithms that mirrored his own color-mixing philosophy.
Work by AARON eventually graduated from laboratory walls to prestigious institutions worldwide, including the Tate Modern and the Whitney Museum of American Art.
The Legacy of 1973 in Today’s AI Era
When we look at modern generative media screens, it is easy to view AI art as a sudden, overnight phenomenon born out of Silicon Valley. But the intellectual DNA of today’s creative algorithms traces directly back to Harold Cohen sitting at a terminal in 1973, wondering if a machine could understand the psychology of a simple line.
AARON proved that art is not strictly bound to the biological hand. By daring to automate the creative process half a century ago, Cohen unlocked a new frontier—proving that the intersection of art and artificial intelligence has a history far deeper, and more human, than we often realize.
Leave a Reply