Shaping new languages of expression with AI.

We have to talk about AI. Untold billions of pixels have been spent, even wasted, talking about AI already. We know the topic feels worn down to the nub. And yet, so far AI has only peeked over the barricade into our world of letters.
Type design and typography have managed to stay just out of the spotlight despite being occasionally drawn into broader conversations about graphic design, “creativity,” and the unsettled question of AI’s role in all of it.
But for better or for worse, it turns out that AI is fairly bad at drawing letters. In a recent piece in The Atlantic (a piece that, it should be noted, was sponsored by Google), writer Drew Campbell observed that while AI knows many words and may already know all the words there are to know, it “stumbles” when “creating the literal letterforms that construct each syllable, clause, and paragraph.” The new OpenAI release, which was made public at the end of March, does seem pretty good at creating letters in context.
“To recreate the work that designers have been doing since the days of illuminated manuscripts and the Gutenberg Bible is no simple feat,” Campbell rightly points out. “At this stage in the evolution of AI tools, the models built on statistical probabilities struggle to recreate the simple beauty of well-designed, meticulously crafted typography.”
Uniqode by Koto. From IRL to URL and back again.
Uniqode by Koto. From IRL to URL and back again.
Øyedrops created a visual motion and design language for Kongsberg Digital and Microsoft — An AI initiative shaping the future of work.
The operative phrase there, of course, is “at this stage.” The true menace of AI is not so much what it’s capable of doing today, but rather that it doesn’t rest. An AI program can learn, evolve, iterate, and work 24/7. Give it a job and it will toil sleeplessly until the task is complete. That’s a wonderful advantage if the job is, say, parsing cancer research data, but less so if it’s practicing and refining a creative craft that threatens to eventually put thousands of people out of work.
“Whilst concerns over the future of the graphic designer persist, the World Economic Forum paints a more positive picture for the broader design industry, especially in UX and UI design roles.”
World Economic Forum Future of Jobs Report 2025
Which brings us around to the inevitable, crucial, and confounding question at the heart of this whole AI conversation: Who, exactly, is steering this ship?
AI Logo Maker by Picsart
Virtually every existing criticism of AI as a whole also applies to the technology’s potential uses in type. Since AI can only regurgitate and remix ideas based on its training data, it can’t create anything new, unique, or exciting. It lacks vision, audacity, and heart.
But as any student of type history knows, our industry has undergone multiple wholesale shifts in how type is made and put to print. Entire printing models have risen to and fallen from prominence in the past century alone, speedwalking us to our current digital era and the unprecedented abundance of typefaces available to designers and brands. We are no stranger to disruption and rapid evolution. But this moment is different. AI is not a faster printing press. It’s not a floppy disk replacing boxes of lead type. AI has the potential to learn the craft — adequately, if not masterfully or soulfully. Surely some revelation is at hand. And so, we sit at this fork in the road, one direction seemingly meandering toward an unsettling precipice and the other twisting and turning toward an uncertain partnership.
ACID the UK’s leading design and intellectual property (IP) campaigning organisation.
“Whether the work is done by our team or others in the industry, we believe human beings should remain central to typographic ideation.”
Monotype
As we explored this topic for Re:Vision, we thought a lot about the responsibility Monotype must assume as AI matures. We are caretakers of more font IP than probably any other company in the world. In AI terms, that’s “training data.” And so we put to print that we are dedicated to protecting the humanity of high-quality type. Whether the work is done by our team or others in the industry, we believe human beings should remain central to typographic ideation. We’ve also open-sourced technology to support font files in the Content Authenticity Initiative’s C2PA coalition, which aims to establish and promote standards for verifying and trusting online content.
That’s all well and good but it will take more than pledges and isolated acts to influence this truly global phenomenon. Someone has to grab the wheel.
Each section of Re:Vision has a companion activation, essentially a brief we’ve created for ourselves and some partners aimed at actually doing something about the issue at hand. In the case of Human Types, we’ve launched three projects simultaneously that aim to explore, challenge, and test the limits of AI as it relates to letters.
We’ve partnered with our friends Matthieu Salvaggio from Blaze Type and Matteo Bologna from muccaTypo, along with Sina Otto from the Monotype Studio and given each a project of their own.
Matthieu will explore AI type design tools in collaboration with Monotype. Our activation asks him to supply our AI team with six publicly available font designs, but that in this instance are reduced in character scope. Our AI output will create six unique AI fonts for his and our review.
Sina will explore how designers are grappling with AI image making tools such as Stable Diffusion, Midjourney, Firefly, etc., to create and embellish letterforms with greater meaning and purpose via decorative artifice.
Matteo will imagine a future where artificial intelligence acts as a personal assistant that truly understands you — even your favorite fonts. In this future, the fonts you see while reading won’t be the same for everyone; they’ll change to match your personality, mood, and needs.
Each of these projects places humans at the center of the exercise — as the creator, creative user, and audience. We hope to show that AI can sit alongside us as a partner (or copilot, if you will) and demonstrate that its value should not be measured by what it replaces, but what it adds.
Explore
Type Trends 2025. The latest in type design, from the Monotype Studio.