Before the Prompts: The Fascinating Story of the First-Ever AI Image
We live in an era where generating a hyper-realistic photograph, a surreal oil painting, or a 3D render takes mere seconds. You simply type a few words into a text box, hit enter, and watch modern algorithms conjure visual masterpieces.
However, the journey of artificial intelligence creating imagery didn’t begin with deep-learning neural networks or multi-billion-parameter models in the 2020s. It started more than half a century ago, born from the mind of an artist who wanted to answer a radical question: Can a machine learn to understand the rules of art and create on its own?
Meet AARON: The Pioneer of Algorithmic Art
The undisputed ancestor of modern AI image generation is AARON, a computer program developed starting in the late 1960s and early 1970s by Harold Cohen, a painter and professor at the University of California, San Diego.
Unlike today’s AI models—which scrape billions of images from the internet and use statistical associations to guess pixels—AARON was built entirely differently. Cohen didn’t want a tool that simply copied human drawings; he wanted to build an autonomous system that possessed its own understanding of visual representation.
By 1973, AARON was officially producing its first autonomous images.
How AARON Worked:
Instead of relying on neural networks (which didn’t exist in a practical form for art back then), Cohen manually coded the “rules” of art into the software. He taught AARON how to conceptualize space, overlapping objects, human anatomy, and composition.
From Room-Sized Computers to Physical Pens
Operating AARON was nothing like opening a web browser today. In the 1970s, Cohen worked with room-sized mainframe computers.
- The Code: Cohen fed instructions and rules into the computer.
- The Decision: Based on those coded rules, AARON would make independent decisions about composition, geometry, and placement.
- The Execution: The computer was hooked up to a mechanical drawing machine (a flatbed plotter) holding a physical pen. The machine would slowly glide across a sheet of paper, moving back and forth to sketch out original black-and-white lines.
The resulting images were abstract, minimalist, and often featured primitive human figures, flora, and geometric shapes. Yet, they represented a seismic shift in human history: the moment a machine was no longer just a passive mirror reflecting human input, but an active participant in creation.
The Evolution: How We Got Here
To truly appreciate AARON’s breakthrough, it helps to look at how AI image generation evolved across generations:
| Era | Technology | How It Worked | What It Produced |
| :— | :— | :— | :— |
| 1960s–1970s | Rule-Based Systems (AARON) | Hand-coded artistic rules, logic algorithms, and physical plotters. | Black-and-white line drawings of figures and abstract shapes. |
| 2010s | Generative Adversarial Networks (GANs) | Two neural networks (a generator and a discriminator) competing against each other. | Synthetic human faces, low-resolution artwork, and style-transferred photos. |
| 2020s–Present | Diffusion Models & Transformers | Text-to-image models trained on billions of parameters tracking text-pixel relationships. | Hyper-realistic photographs, complex paintings, and cinematic renders instantly. |
A Lasting Legacy
Harold Cohen spent decades updating AARON, eventually teaching it to handle color, texture, and increasingly complex jungle-like scenes. His creation went on to be exhibited in prestigious spaces like the Tate Modern and the Whitney Museum of American Art.
Leave a Reply