The Canvas in the Machine: How 1973 Sparked the Era of AI-Generated Art

When we think of artificial intelligence generating images, our minds usually jump to modern marvels: sleek digital prompts, hyper-realistic portraits, and sprawling fantasy landscapes conjured in seconds. But long before high-powered neural networks and internet datasets existed, the groundwork for machine-made art was laid over half a century ago.

The story of the world’s first AI-generated image takes us back to 1973—and to a visionary British painter who decided to teach a computer how to draw.

Meet Harold Cohen and AARON

In the early 1970s, Harold Cohen was a successful, internationally exhibited painter representing Britain at prestigious events like the Venice Biennale. Yet, he found himself gripped by a radical question: What is the core essence of drawing? To find out, he stepped away from conventional fine art and walked into the artificial intelligence laboratory at Stanford University.

Cohen wasn’t a computer scientist by training, but he understood the logical architecture of visual art. Working at Stanford, he began developing a groundbreaking software program that would change visual culture forever: AARON.

Unlike modern AI tools that scour billions of images using machine learning, AARON was built on symbolic AI and hand-coded rules. Cohen didn’t train it on a database of existing artwork; instead, he manually translated his own understanding of cognitive and visual rules into computer code.

He taught the program concepts like:

• The distinction between figure and ground
• How objects overlap in space
• Basic rules of composition and boundary control

The 1973 Breakthrough: Art Without Human Intervention

By 1973, Cohen’s efforts culminated in a monumental milestone. Running on early computing hardware, AARON began generating abstract, intricate compositions of lines and closed forms on paper using a connected physical plotting machine.

What made these 1973 outputs historically significant wasn’t just that a computer drew them—it was autonomy. Once Cohen pressed start, he did not guide the mechanical arm or intervene in the creative process. The machine made its own real-time decisions about composition, layout, and line weight based on its internal rules.

For the first time in history, an autonomous system conceptualized and executed a visual artwork independently of its creator’s hand.

From Black-and-White Lines to a Lifelong Partnership

The earliest 1973 iterations of AARON produced relatively simple, abstract, black-and-white line drawings. However, Cohen didn’t view AARON as a one-time experiment. He spent the next four decades continuously updating the software.

As computing technology evolved, so did AARON:

• In the 1980s and 90s, Cohen expanded AARON’s rulebook to understand color, shading, and texture, eventually allowing it to paint complex organic shapes, rocks, plants, and human figures.
• AARON’s works graduated from laboratory printouts to being featured in world-renowned institutions like the Tate Modern and the Whitney Museum of American Art.

Why 1973 Matters Today

Looking at AARON’s primitive 1973 line drawings next to today’s hyper-detailed digital generators feels like comparing the Wright brothers’ flyer to a modern supersonic jet. Yet, AARON proved a profound philosophical concept decades before its time: creativity was not strictly an exclusive biological trait.

Harold Cohen and AARON opened the door to a completely new medium, shifting art from a purely physical act to a conceptual collaboration between human intent and machine execution. The next time you type a prompt into an AI image generator, remember that the spark of that technology was struck back in 1973—not by engineers trying to build a business, but by a painter trying to figure out how we see.