No shortcuts: why AI threatens creative instinct

A tool is only essential so long as nothing more convenient exists. What survives is the judgement it was built to partner with.

Consider the California job case: a wooden tray that organises the moveable metal letters that slot into a letterpress. Systemised and deceptively complex like a city gridiron, each box neatly houses a family of letters. A typesetter would spend years learning to navigate the case fluently: to pull type by touch fast enough to keep pace with a running press. It was the organisational backbone behind mechanised printing, which put text in front of more people, faster than any scribe could. So foundational to the practice of typography that we still call the capital letters ‘uppercase’ and the small ones ‘lowercase’.

Today, the California job case is a novelty item. If you’re lucky enough (and share my sense of what constitutes luck), you can find one at the odd estate sale. While using one was never part of my practice, I keep one in my studio to remember that our tools are always provisional, and that the part of the work worth holding onto was never the speed of the hand.

Typographic panic

The turnover of tools that define a practice can feel disruptive. When page layout software like Adobe (and QuarkXPress before it) swept through design studios, with it came the fear that a door closed every time an Adobe program opened. Designers feared the new technology would eradicate skills so carefully cultivated through painstaking iteration, and the creative judgement developed in parallel would dissipate. The field adapted anyway; tools like Adobe and Figma became the baseline, and the debate shifted to what practitioners needed to learn as the tools changed.

Designers have been telling themselves similar stories about artificial intelligence. Some have revived prophecies of discipline collapse, while others dismiss the panic entirely and assure peers that the field will adapt as it always has. There is some truth in the latter notion, but there’s a distinct nuance to today’s upheaval that’s worth teasing out.

Beyond imitation

AI reaches into a part of the job the previous upheavals left alone. It can handle the entire process of creation from end to end, and if used without discretion, can ship an end product with minimal human participation. Scope has grown, too, since these tools now touch writing, code, and strategy alongside visual design. But more significantly, automation has moved from execution to ideation, and no earlier tool in the field’s history did that. Judgement is set to overtake technique as the practitioner’s most important contribution, but it has never been easier to reach a final product without it.

How we frame our relationship to that sort of technology matters: treating it as an oracle to consult, a replacement to install, or a collaborator to work alongside leads to very different outcomes for who ends up with authorship over the work.

Economist Erik Brynjolfsson has a useful framework to focus on the fear of AI adoption: the Turing Trap. The name references Alan Turing, who proposed in 1950 that a machine could be judged intelligent if a person conversing with it could not tell it apart from another human. Brynjolfsson rightly points out that the sole goal of imitating a person is misguided. It aims a technology capable of replicating human output at basic substitution, rather than complementing or extending what a person can do.

The case for augmentation over substitution is clear in design, both for the sake of the work and the people behind it.

First, let’s consider the work. Design optimised for AI scraping is worth its own discussion, but humans nevertheless remain a priority audience. Work that passes through a human understanding of how design makes a person feel – and that draws on the creative legacy of generations who’ve honed the craft of communicating information and sentiment – will keep its edge.

Now, consider the practitioner. Underneath every argument about tools and technique sits a question about labour: namely, who gets to do creative work, on what terms, and whether that work still holds meaning for the people doing it. Unfortunately, many designers have accepted unstable hours, modest pay, and long apprenticeships precisely so they can keep exercising their creativity. Those terms were not chosen by any one designer; they are often conditions of the field.

Creative labour is one of the main ways people find meaning in their working lives, and that meaning isn’t only in the finished product.

Teaching taste

Junior designers used to refine their judgement by executing someone else's direction, each iteration the product of veteran judgement, before they earned a say in the bigger decisions. The sum of entry-level tasks was always more than the sum of their parts: that iterative process allowed junior practitioners to build an instinct for defensible design decisions. Struggle isn't worth romanticising for its own sake, but its role as an intermediary between novice and master historically played a considerable role in building taste.

Eliminating this rote work isn’t inherently bad, so long as you compensate for the muscle it built.

Professors and practitioners must continue to drill in the importance of critical thinking and creative history: passing down the frameworks that our predecessors spent lifetimes developing, and challenging our greener counterparts to incorporate it, to question it, and to ultimately use it to make sharper design choices that connect with their audience.

Kathy Pham recently cautioned against reducing design education to a trade school model; that is, prioritising technical training at the expense of the liberal arts foundation that produces critical thinkers. In a design scene dominated by AI, a curriculum that teaches only technique trains students in the skill they can least rely on.

As an academic myself, I see a number of ways educators can lean into cultivating judgement as the premium on technique lessens.

History and theory should be prioritised in order to give students a wider, systems-level understanding of the ideas in their work.

The volume of work and scope of projects should be scaled to allow students to spend more time inside the work, rather than forcing students to hand off tasks to AI just to keep up. With this, assessment can move towards the record of how a student reached a decision rather than the polished final image – placing a premium on process over product.

Critique, the studio tradition of dismantling and defending work in front of peers, will take on newfound importance as students are pushed to unpack their reasoning. The studio can be taught as a place where people make decisions together.

The private sector stands to gain from this focus as well, and thoughtful adopters are already thinking and acting in kind. As stated in a recent Figma report, "when prototyping is fast, choosing what ships is more important… Decisions get sharper when teams work through them together: weighing the conflicts to consider, the directions to explore, and which trade-offs to make.” That is true inside a studio, and it is true of the field as a whole. The choices that shape how these tools enter creative work are not made in silos.

Costs beyond craft

As my colleagues and I have emphasised at the Lab for AI, Ethics, and Creative Labor, using AI well largely comes down to protecting human judgement and agency.

Intellectual property is one major issue at play: these models learned from the work of generations of designers and creators without asking permission or offering payment, and that same body of work now competes against the people who made it.

Concentration of power is another concern: a small number of companies own the models and the computing behind them, and get to set the terms under which everyone else practices the craft. Designers deserve to retain some say over the conditions of their own craft, including a seat at the table when these tools are built.

None of that is won by individuals. Judgement is the practitioner’s to build, but the conditions under which anyone gets to keep building it are secured collectively or not at all. Writers, filmmakers, and dubbing artists are already organising on exactly these terms, through unions, guilds, and cross-field alliances, and communication design has been slower to join them than its stakes warrant. Part of what these alliances offer is a clear line of sight into neighbouring fields living through the same shift in real time, and the Lab’s own Coalition for AI and Creative Labor is one attempt to build it.

AI generation also carries a real environmental footprint, and these systems inherit the errors and biases embedded in their training data – so a tool asked to generate ‘professional’ imagery or ‘neutral’ copy is not guaranteed to produce without fault. Even if the space here is too short to do them full justice, these issues belong alongside the judgement question because they raise the stakes for the preservation of human discernment.

Tools don’t decide

In spite of criticism, AI tools and processes are now permanent fixtures of the design world – at least until something more effective and convenient comes along. What designers need to build now is the instinct to push back; to keep deciding what belongs on the page, on the screen, or in the experience.

The California job case still sits in my studio, a reminder that nobody mourns the compositor’s carpal tunnel, because the judgement that hand-speed served survived without it. Whether the same holds for the next generation depends on choices professors and practitioners make right now. The technology cannot decide for them.