For all the hand-wringing over artificial intelligence, the truth is, it’s already here. Fire up InDesign or Photoshop, and little moments of machine learning will help you along the way. At this point, AI’s practical uses are mostly subtle, relegated to tools like content autofill and auto brightness, but what happens when AI takes a more creative role in graphic design?
It’s something Barney McCann has been thinking about lately. The designer and creative director has spent the last year exploring how AI can create new forms in typography, and the result is a series of wild, amorphous lettering that looks like it’s been chewed up and spat out of a computer’s brain.
At its most basic, McCann’s Artificial Intelligence Typography project explores what happens when AI tries to transition from one letter to the next most efficiently. Using Arial as a starting point, McCann fed letterforms into a neural network that analyzed the basic shape of each letter. The algorithm learned the slant of a capital A and double roundness of a capital B; it noted the simplicity of the letter I and the complexity of a G. Then McCann told it to write.
As A transforms to B and B to C, the computer cycles through a countless number of variations in order to find the most efficient way to produce the next letterform. It understands, for example, that an L consists of two straight forms that intersect at a right angle in the lower left hand quadrant of a grid. It also understands that an I is a straight vertical line in the center of the grid. Moving from one to the other requires a series of steps and shifts that McCann wanted to make visible. “I suppose the idea is to create a typographic language that can create coherent and legible sentences using as few shapes as possible,” he says.
I think this is the closest you can get to a computer’s handwriting
This transition happens all the time when we type, he explains, only we don’t see it. Pixels on screen shape shift imperceptibly. When you type “Eye on Design,” you can’t see the transition from E to Y, but it happens nonetheless. “What you normally see is the beginning and end—the computer’s first step and last step,” McCann says. “This is showing the whole process.”
Using the coding language of Processing, McCann turned the messy process of that transition into a shape-shifting typeface that he’s tentatively calling Obsolete. “It’s a semi-sarcastic reference to the idea that it will make designers obsolete,” he says.
Obsolete’s letterforms are inherently unpolished. It’s almost as if you’re watching typographic genetics at work—as one letter transforms to the next, you can see traces of both in the final form that emerges. Obsolete’s letters are products of their surroundings. Everything from the sequence of letters in a word to the speed of a computer and a wifi network can affect its final shape. “It’s the same as nature or nurture,” McCann says. “The letters are the same by nature, but their nurture is completely different.”
It’s almost as if you’re watching typographic genetics at work
As a moving typeface, Obsolete feels like a distant cousin of the parametric type movement, which use algorithms and pre-set parameters to create shape-shifting generative typefaces. Like a parametric font, Obsolete operates within the boundaries of rules, but those rules are loosely held enough to allow for relatively unstructured forms to emerge. “I think this is the closest you can get to a computer’s handwriting,” he says.
For McCann, Obsolete represents a more creative approach to the notion that AI is increasingly influential in our design decisions. “AI is already making decisions for you,” he says. “You’re just approving decisions now.” In some ways, the typeface peels back the opaque veneer of mystery around AI. It reveals a computational thought process that McCann believes is, at the very least, could be used as a jumping off point for new ideas and forms. “It’s a good starting point,” he says. “But this is still an unfinished experiment.”