Skip
Current Issue
This Month's Print Issue

Follow Fast Company

We’ll come to you.

1 minute read

What Happens When You Apply Machine Learning To Logo Design

This web bot will generate millions of logos until you find the perfect one.

  • 01 /07
  • 02 /07
  • 03 /07
  • 04 /07
  • 05 /07
  • 06 /07
  • 07 /07

The rise of neural networks and generative design have created new opportunities for designers. But what if it went the other way, and robots created a Skynet that kills off human designers (or at least their careers) once and for all?

A heady question. Depending on whether you embrace or fear the robo-future of design, Mark Maker (via Sidebar) could be considered either the beginning of the end, or proof that such fears are overstated, because bots are still pretty crap at design. Either way, it's a fun web toy.

In Mark Maker, you type in a word. The system then uses a genetic algorithm—a kind of program that mimics natural selection—to generate an endless succession of logos. When you like a logo, you click a heart, which tells the system to generate more logos like it. By liking enough logos, the idea is that Mark Maker can eventually generate one that suits your needs, without ever employing a human designer.

Mark Maker creates its logos by breaking each design in half, so that it contains both a base design and an accent element. The example Mark Maker's creators use is the Mobil logo, which contains a blue sans serif typeface as the base, and a red "o" as an accent. The typefaces are plucked from Google Fonts' open-source typeface library, while icons come from the Noun Project.

To be honest, I found myself grudgingly impressed by some of the logos Mark Maker produced for us. In fact, when I entered "Co.Design" into Mark Maker, the program generated some marks that looked eerily similar to Co.Design's own logo, which was designed in-house. Or was it?

Try Mark Maker for yourself here.

loading