What Was the First AI-Generated Image in History?

Long before text-to-image prompts could summon hyper-realistic portraits in seconds, the world’s first AI-generated image was born out of ink, mainframe computers, and a radical artistic experiment.

The story of the first artificial intelligence image is not about a neural network scraping the internet. It is about a British painter who decided to teach a machine how to think like an artist.

Meet AARON: The World’s First AI Artist

In the late 1960s, Harold Cohen, a successful British-born painter and professor at the University of California, San Diego, grew deeply fascinated by the cognitive rules behind drawing. He wondered: What are the minimum conditions under which a set of marks on a page functions as an image?

Rather than just using a computer as a passive tool, Cohen wanted the machine to make autonomous creative decisions. Beginning in 1968, he set out to write a groundbreaking software program that would eventually be called AARON.

By the early 1970s—specifically taking shape through 1972 and 1973—AARON began producing its first autonomous black-and-white line drawings. Running on massive, refrigerator-sized minicomputers, AARON didn’t rely on “machine learning” or training data like modern models do today. Instead, it relied on Good Old-Fashioned AI (GOFAI): a massive, hand-coded set of symbolic rules.

Cohen painstakingly translated his own understanding of geometry, perspective, occlusion, and human anatomy into code. AARON was taught how to balance proportions, construct shapes, and arrange compositions on its own.

[Harold Cohen's Mind / Art Rules] 
       │
       ▼
[AARON Software (1970s)] ──► Makes Autonomous Decisions ──► [Plotter / Robotic Arm draws lines]

From Code to Canvas: How It Was Made

Because computer screens in the early 1970s were crude and incapable of displaying fine art, AARON didn’t output digital pixels to a monitor. Instead, the software outputted coordinates and commands to a flatbed plotter—a mechanical drawing machine equipped with a physical pen.

As the program executed its internal rules, the mechanical arm dragged across paper, drawing crisp, abstract figures, rocks, plants, and rudimentary human forms line by line.

This made AARON a true pioneer:

• Autonomous Composition: While other computer artists of the 1960s (like Georg Nees or Frieder Nake) wrote algorithms to generate precise geometric patterns, AARON made independent structural and compositional choices within its programmed boundaries.
• The Studio Assistant Dynamic: Cohen famously compared his relationship with AARON to that of a Renaissance master and a studio assistant. He wrote the rules and set the parameters, but the machine chose how the final image was constructed.

Evolution: From Line Drawings to Full Color

AARON didn’t stay static in the 1970s. Over a multi-decade collaboration that lasted until Cohen’s passing in 2016, the software evolved dramatically:

1. The 1980s & 90s: Cohen taught AARON how to handle closed shapes, allowing the system to tackle complex figurative scenes featuring people and lush botanical environments. By the mid-1990s, AARON was outfitted with robotic brush-and-water systems that allowed it to pick up paint and color its own creations autonomously.
2. Museum Recognition: AARON’s work transitioned from computer science labs to prestigious cultural institutions, including the Tate Modern, the Whitney Museum of American Art, and the San Francisco Museum of Modern Art.

Why the First AI Image Matters Today

Looking back from an era where generative models synthesize photorealistic worlds in a fraction of a second, AARON’s shaky black-and-white line drawings might look quaint. Yet, AARON fundamentally shifted human philosophy.

It proved that the line between human expression and machine execution was not absolute. Harold Cohen didn’t use a machine to copy an image; he successfully encoded artistic reasoning into software for the very first time, cementing AARON as the undisputed ancestor of modern generative AI.