The History of the First AI-Generated Image Ever Created

Beyond the Prompt: The Surprising History of the First AI-Generated Image

When we think of artificial intelligence and images today, our minds instantly race to hyper-realistic text-to-image generators. With just a few descriptive words typed into a prompt bar, modern systems can conjure up cinematic photos, surrealist paintings, or intricate digital illustrations in mere seconds.

Yet, the story of how machines first learned to draw doesn’t begin in a high-tech Silicon Valley lab with neural networks or vast training datasets. It begins in the late 1960s, inside the mind of a bewildered British painter who asked a radical question: What if a machine could understand the cognitive rules of art?

Meet AARON: The World’s First AI Artist

Long before deep learning or diffusion models, a professor and artist named Harold Cohen started building a computer program that would change creative history. In 1968, while working at the University of California, San Diego, Cohen—a successful painter representing Britain at prestigious global art events—grew deeply fascinated by computing. He didn’t want to use a computer merely as a digital paintbrush; he wanted the computer to make decisions on its own.

By 1973, Cohen’s creation—an autonomous software system named AARON—successfully produced its first true AI-generated images.

Unlike modern AI, which scrapes billions of images from the internet to recognize statistical patterns, AARON operated on pure logic. Cohen didn’t feed AARON pictures to copy. Instead, he painstakingly hand-coded human artistic rules into the program:

• How objects overlap in space.
• How to balance proportions and compositions.
• How to structurally represent figures, shapes, and eventually botanical elements.

“AARON was not a tool that executed commands; it was an autonomous entity programmed to exercise creative judgment within boundaries defined by its creator.”

What Did the First AI Images Look Like?

If you were expecting a masterpiece of color and light, the reality is wonderfully primitive. AARON’s earliest outputs were stark, black-and-white line drawings.

Running on room-sized computers of the era, AARON would drive specialized mechanical drawing machines—plotting pens gripped by robotic arms—across large sheets of paper. The resulting artwork featured abstract human figures, bold geometric shapes, and organic forms.

Crucially, AARON never knew what it was drawing. It didn’t possess a “concept” of a human or a plant the way humans do; it understood spatial constraints, boundaries, and relational rules. Yet, because the program made randomized choices within its coded ruleset, no two drawings were ever identical. AARON was improvising.

From Mechanical Pens to Neural Networks

As computing power evolved over the decades, so did Cohen’s digital apprentice.

• In the 1980s and 90s, Cohen updated AARON to understand color theory, teaching the program how to mix its own paints and fill in the shapes it drew.
• AARON’s works graduated from laboratory curiosities to celebrated exhibitions displayed at major institutions, including the Tate Modern and the Whitney Museum of American Art.

Cohen passed away in 2016, but AARON laid a monumental philosophical and technical bedrock. It proved that creativity wasn’t a sacred, strictly biological trait exclusive to humans—it could be broken down, codified, and mirrored by logic.

Why the First AI Image Still Matters

Today, as we navigate a world saturated by instant AI-generated media, looking back at AARON offers a grounding perspective.