Drawing with pen and paper is one thing, but sketching on a computer is even more difficult for the less visually inclined among us. Simple programs like Paint force you to draw with your mouse or trackpad, which makes doodling even the most basic smiley face a struggle.
This week, Google released a new AI experiment called AutoDraw, which turns your half-baked scribbles into poster-ready clipart. The tool uses machine learning to guess what you’re trying to draw and then gives you the option to replace your bad drawing with more polished ones, created by illustrators and design studios like HAWRAF, Erin Butner, Julia Melograna, Pei Liew, Simone Noronha, Tori Hinn, and Selman Design. It’s a simple tool that gives those of us without fancy (and expensive) design programs a way to make reasonably professional graphics.
In my first attempt to draw something, I started to draw a person. After sketching a blob that distantly resembled a foot, and the program quickly showed me several different types of feet and socks. A crude attempt at a bicycle offered some eyeglasses, motorcycles, a moped, and curiously, two types of pears. I drew a five-point star, and AutoDraw offered me different types of stars, setting suns, boomerangs, lightning bolts, compasses, kitchen appliances, mirrors, and even a paper clip. Even if you’re truly horrendous at drawing basic shapes, chances are you’ll be able to find what you’re looking for.
The end of creativity? Maybe. But AutoDraw is more than just a way to easily input clipart. It illustrates how Google’s AI experiments are evolving. The web app is actually based on the technology used in the company’s Quick, Draw! experiment from 2016, which used machine learning to guess what you were doodling. Paired with work from human designers, AutoDraw shows how this kind of AI can be used in the real world.