The Ghost in the Machine: The Untold Story of the First-Ever AI Image

When we think of artificial intelligence and imagery today, we picture instant text prompts, ultra-realistic digital renderings, and fierce online debates over copyright and creativity. Tools like Midjourney, DALL-E, and Stable Diffusion allow anyone to conjure cinematic masterpieces in seconds.

Yet, the lineage of AI-generated art stretches back much further than the 2020s. Long before neural networks were trained on billions of scraped photographs, the first true AI image was born out of an artist’s profound existential question: Can a machine learn to draw on its own?

This is the story of AARON, the world’s first AI artist, and the origin of artificial intelligence imagery.

The Man Who Left the Art World Behind

To understand the first AI image, you have to look back at the late 1960s. Harold Cohen was a celebrated British painter. He had represented Great Britain at the prestigious Venice Biennale and exhibited his abstract canvases in world-class galleries. By traditional metrics, he had made it.

Yet, Cohen grew restless. He began wondering whether the traditional definitions of art-making were too narrow. Crucially, he became fascinated by computers—not merely as digital tools to edit or scan existing work, but as entities capable of independent cognition.

In 1968, Cohen abandoned his thriving London career, moved to California, and joined the faculty at the University of California, San Diego. There, he taught himself computer programming. He didn’t want to use a computer to automate brushstrokes; he wanted to codify the cognitive processes behind why humans make marks on a page.

Enter AARON: The Pioneer of Autonomous Art

By 1973, Cohen’s experiment materialized into a computer program that would change art history: AARON.

Unlike modern AI, which relies on statistical deep learning, neural networks, and massive image datasets, AARON was built entirely on symbolic artificial intelligence and manually coded rule sets. Cohen spent decades systematically breaking down the rules of visual cognition. He programmed AARON to understand:

• The physics of drawing: How lines intersect, close, and form shapes.
• Spatial reasoning: How to compose elements on a page without overlapping awkwardly.
• Human and botanical forms: Abstracted rules for generating figures, rocks, and foliage.

When AARON generated its first primitive, black-and-white line drawings in the early 1970s (publicly debuted at UC Berkeley in 1974), it wasn’t copying a human drawing or stitching together a collage. Connected to a custom-built mechanical drawing machine, the computer made its own compositional decisions in real-time. Cohen would supply the program with parameters, but he never knew exactly how a given drawing would turn out. The machine generated original compositions within the boundaries of its programmed logic.

“AARON was perhaps the longest-lived and certainly the most creative artificial intelligence program in daily use.” — Dr. Paul Cohen

A Lifelong Collaboration

Over the next four decades, AARON evolved remarkably. What began as abstract, minimalist line drawings reminiscent of primitive petroglyphs gradually grew in sophistication.

• The 1980s and 90s: Cohen taught AARON how to handle color, with the program painting intricate, vibrant shapes while Cohen carefully mixed the pigments it requested.
• The 2000s: AARON advanced to painting complex figurative scenes featuring human-like silhouettes interacting with lush, botanical environments.

Institutions like the Tate Modern and the Whitney Museum of American Art began exhibiting AARON’s work, acknowledging its historical weight. Cohen and AARON maintained a collaborative partnership until Cohen’s passing in 2016.

Why the First AI Image Still Matters Today

Looking at AARON’s jagged, minimalist 1973 drawings next to today’s hyper-polished AI generations feels like comparing a Wright brothers’ airplane to a modern supersonic jet. Yet, the fundamental philosophical question AARON posed remains entirely unchanged.

Harold Cohen proved that creativity isn’t an exclusive, mystical human biological trait—it is deeply rooted in systems, rules, and autonomous decision-making. Long before the era of generative neural networks, AARON showed us that a machine could hold a brush, look at a blank canvas, and create something that never existed before.