What Was the First AI-Generated Image? History & Origin

The Ghost in the Machine: What Was the First AI-Generated Image?

When we think of AI-generated art today, our minds instantly flash to hyper-realistic digital paintings, surreal landscapes rendered in seconds, or viral text-to-image prompts typed into tools like Midjourney or DALL-E. It feels like a thoroughly modern phenomenon—a technological magic trick born in the 2020s.

Yet, the lineage of artificial intelligence creating art stretches back far further than you might think. Long before neural networks, diffusion models, and massive internet scrapers, a lone British painter asked a radical question: Can a machine be taught to draw autonomously?

The answer resulted in the world’s first true AI-generated images, created by a pioneering software program named AARON.

Meet Harold Cohen and AARON (1973)

In the late 1960s, Harold Cohen was a celebrated, internationally recognized painter. He had represented Great Britain at elite global exhibitions like the Venice Biennale. But at the peak of his traditional career, Cohen grew restless. Fascinated by the structural logic of art, he left the established art scene in London to take a teaching position at the University of California, San Diego, where he taught himself how to program computers.

Cohen wasn’t interested in using a computer merely as a digital paintbrush or a fancy copy machine. He wanted to externalize the cognitive process of drawing. He began writing a computer program that would become known as AARON.

By 1973, AARON produced its first breakthrough autonomous images.

How the First AI “Thought”

Unlike modern AI, which relies on machine learning algorithms trained on billions of images scraped from the web, AARON had zero access to a dataset or internet database. It didn’t “look” at photos to learn how to replicate reality.

Instead, AARON was built entirely on symbolic logic and rules-based programming. Cohen meticulously coded his own understanding of art into the software:

• Rules about how lines intersect.
• The geometry of open and closed shapes.
• Principles of spatial composition, balance, and proportions.
• Physical rules regarding how objects overlap in a scene.

When AARON ran, it made independent decisions. Cohen didn’t pre-draw the images or dictate every stroke; he provided the cognitive framework, and the software dynamically composed unique line drawings on the fly. Connected to a mechanical drawing machine (a robotic flatbed plotter with a pen gripped in its arm), AARON would physically execute the artwork on paper all by itself.

“Cohen found that some of his programming instructions generated forms he hadn’t imagined. The machine was showing him possibilities within his own artistic system that he hadn’t seen.”

The Evolution of the World’s First AI Artist

The earliest iterations of AARON generated primitive, abstract black-and-white line art that resembled ancient petroglyphs or minimalist geometric shapes. But AARON was a life’s work. Over the next four decades, Cohen continuously updated the program:

• The 1980s–1990s: AARON was upgraded to handle color, with the program calculating its own color palettes and mixing liquid paints using specialized robotic machinery.
• The 2000s: The software evolved past pure abstraction to successfully generate complex figurative shapes, depicting human figures, stylized plants, and eventually rich, jungle-like environments.

AARON’s work became so culturally and historically significant that it was exhibited at prestigious venues worldwide, including the Tate Modern and the Whitney Museum of American Art.

Why the First AI Image Still Matters Today

When Harold Cohen passed away in 2016, his partnership with AARON ended, but their combined legacy permanently altered the horizon of human creativity.

Looking back, AARON reminds us that the debate surrounding AI and art is not actually new. Fifty years ago, the art world was already grappling with fundamental questions: Who is the true author of an algorithmic creation—the programmer or the program? Can a machine possess genuine creativity?

While today’s AI models operate on a vastly different technological scale—transforming text prompts into hyper-detailed art via neural networks—they all share a philosophical ancestor. That ancestor is a 1970s mainframe computer humming away in California, guided by an artist who simply wanted to see what would happen if he taught a machine how to think about a line.