The Ghost in the Machine: A Brief, Strange History of Early AI Pictures

We live in an era where typing a casual sentence into a text box summons hyper-realistic, breathtaking digital paintings in seconds. It feels like magic. Yet, the urge to teach machines how to look, interpret, and render images is decades older than modern smartphones, web browsers, or even personal computers.

Long before text-to-image models sparked global conversations, early pioneers were coaxing primitive computers and early neural networks into creating visual art. These “early AI pictures” weren’t born from instantaneous cloud-based generators; they were the hard-fought result of plotters, rule-based logic, and computerized fever dreams.

1. The Rule-Bound Pioneers (Late 1960s–1970s)

Before artificial intelligence meant multi-layered deep learning networks, it meant GOFAI—Good Old-Fashioned AI. Researchers were obsessed with symbolic logic and whether a machine could be taught human rules for creative acts.

• The Birth of AARON: Beginning in the late 1960s at the University of California, San Diego, artist and programmer Harold Cohen built AARON, arguably the most famous early automated artist. AARON wasn’t a neural network; it was a robotic drawing machine governed by a massive set of symbolic rules about geometry, composition, and physical movement.
• The Process: AARON would sit at a specially constructed drawing table holding a physical pen, executing code that Cohen wrote. The results were precise, abstract line drawings. What made it revolutionary was that Cohen didn’t pre-draw the final image; he coded the act of drawing, allowing the machine to generate compositions he hadn’t explicitly visualized beforehand.

2. The Era of Machine Pareidolia: DeepDream (2015)

Fast forward a few decades through computational winters and breakthroughs in artificial neural networks. By the mid-2010s, computer scientists had built deep convolutional neural networks designed primarily for object recognition—essentially teaching computers to look at a photo and say, “That is a dog,” or “That is a teacup.”