Essays / Process

Generative Art: When Algorithms Become the Brush

Generative art is artwork made through a system — a set of rules or an algorithm the artist designs and then allows to make some of the decisions. Instead of painting a single image, the artist authors a process, sets its constraints, and lets it produce outcomes that no one could have drawn by hand. The brush, in other words, becomes the code.

This is older than it sounds. Long before computers, artists used rule-based systems — Sol LeWitt's wall-drawing instructions, the chance operations of mid-century composers, the woven logic of textiles. What computation added was speed and scale: a rule that once took a week to execute by hand could now run ten thousand times in a second, each run slightly different. That shift changed not just what art could look like, but what it meant to be its author.

Authoring a system, not an image

The central mental shift in generative art is moving up one level of abstraction. A painter decides where every mark goes. A generative artist decides the rules by which marks are placed, then surrenders the specific outcome to the system. You are no longer choosing the final pixels; you are choosing the space of all possible images and tuning it until the ones that emerge feel right.

You stop asking "what should this image be?" and start asking "what should be true of every image this system can make?"

This is why generative artists talk about "the output space" rather than "the piece." A well-made system has a personality — a recognizable range — even though every instance is unique. Getting that range right is the real craft, and it is far harder than it looks.

Randomness and constraint

The engine of generative art is controlled randomness. Pure randomness is noise; pure determinism is repetition. The interesting work lives in the tension between them, where the artist hands the system enough freedom to surprise but enough structure to stay coherent.

Where randomness helps

• Variation within a family. Slight jitter in position, color, or scale makes a thousand outputs feel related but never identical.
• Emergence. Simple rules applied repeatedly can produce complexity — flocking, growth, erosion — that no one explicitly designed.
• Discovery. Randomness surfaces compositions the artist would never have thought to try.

Where constraint matters more

Left unchecked, randomness collapses into mush. The artist's judgment shows up in the limits: a tight palette, a grid the chaos must respect, a rule that forbids overlap, a curated range of seeds. Many generative artists describe their job as mostly removing possibilities — pruning the system until almost everything it produces is good.

So what does the artist actually do?

A common misreading is that the computer "makes" generative art, leaving the human as a bystander. The reality is the opposite. Every meaningful decision — the concept, the rules, the constraints, the palette, the curation of which outputs to keep — is human. The algorithm executes; the artist authors, judges, and edits. Authorship lives in the system's design and in the taste applied to its results.

It is worth separating this practice from recent AI image generation, which is often confused with it. Most generative art relies on explicit rules the artist writes and fully understands, using tools like Processing, p5.js, or node-based environments. AI models can be used generatively, but they are one tool among many, not the definition of the field. The throughline is older and simpler: an artist building a system, introducing variation, and shaping what emerges.

Done well, generative art offers something handmade work cannot — a glimpse of a whole space of possibilities at once, authored by a person who decided exactly how strange that space was allowed to be.

Iris Lindqvist is a creative coder and generative artist working in p5.js and plotter drawings.