Design Education in the AI Era

I’ve been sitting with a question for a while now: if AI can generate interfaces, imagery, code, prototypes, and increasingly complete digital products, what should design education teach?

‍

I came to this question while I was a professor at SCAD. I remember experimenting with Claude Code during the winter break, and I quickly came to the realization that nearly everything I had been teaching my students would need to be rethought. My graduating seniors, the ones I’d trained to build clickable prototypes in Figma, were about to enter a field where that process was already being replaced by AI tools capable of producing functional code. Higher fidelity, richer feedback, a genuinely different process, not just a faster version of the old one. I had spent many years teaching essentially the same UX curriculum, and I watched it become partially obsolete in the span of a year.

‍

The traditional design degree was built for a world where knowledge, tools, and professional practices changed slowly. A student could learn a workflow in year one and still be using something close to it by graduation. That world I’m afraid is disappearing. A tool a first-year learns today may be obsolete, or transformed beyond recognition, by the time they walk across the stage.

‍

In this new context, design education needs to evolve.

The goal, restated

If tools are moving too fast to anchor a curriculum to, then the curriculum needs a different anchor. I think it’s this:

Develop designers who can keep designing even when the tools, mediums, technologies, and profession itself change around them.

That means splitting design education into two layers. A durable layer of principles and human competencies that change slowly. And an adaptive layer of technologies, tools, methods, and practices that should keep evolving, by design.

First, some first principles

Before getting to structure, it’s worth being explicit about what design actually is, because a lot of curriculum design quietly assumes an answer without saying it out loud.

1. Design is creative problem solving, not self-expression. It exists in response to something outside the designer: a problem, a need, a constraint, a stakeholder.
2. There is rarely one correct design. Designers work with incomplete information and competing constraints. Their job isn’t to discover the answer, it’s to produce the best available solution given what’s known right now, and to keep improving it. Digital design in particular is a sequence of increasingly informed solutions, never a finished object.
3. Design is pragmatic. A beautiful solution that can’t be built, financed, understood, or maintained isn’t a successful design. Desirability, viability, and feasibility have to coexist.
4. Good design is contextual. There’s no universal rubric where originality or minimalism or novelty always wins. A banking interface and an experimental fashion site shouldn’t be judged by the same standard. The real question is always: what matters most here?
5. Originality is a means, not an end. Novelty can create delight and differentiation, but it can also hurt usability. A familiar pattern can be the better design precisely because people already understand it.
6. Craft includes execution. Users don’t experience the designer’s intention, they experience the implementation. A well-conceived idea that’s poorly built becomes a poorly experienced product. Craft is the fidelity between the two.
7. Design judgment has to stay connected to material reality. For digital designers, that material is now software, data, APIs, AI models, and computation. Not having to code isn’t the same as not needing to understand technology. Designers need enough technical understanding to know what’s possible, what’s expensive, and what’s still unexplored, so they can push past convention without drifting into fantasy.

Craft is the fidelity between intention and reality.

This isn’t a new idea. It’s close to a hundred years old. When Walter Gropius opened the Bauhaus in 1919, he built the entire school around collapsing the wall between craft and creative vision, insisting that the artist was, at bottom, simply a craftsman working at an exalted level, with no essential separation between the two. Every student, regardless of discipline, learned material, technique, and construction before they were trusted with form. The tools have changed completely since 1919. The underlying claim, that judgment has to be grounded in an honest understanding of what you’re actually building with, hasn’t aged a day.

‍

Dieter Rams took that same instinct and turned it into a working checklist decades later at Braun. His ten principles of good design are worth sitting with, not because they should be gospel, but because they’re a rare example of principles durable enough to survive a total change of medium, from radios and razors to the software most of us design today:

• Good design is innovative
• Good design makes a product useful
• Good design is aesthetic
• Good design makes a product understandable
• Good design is unobtrusive
• Good design is honest
• Good design is long-lasting
• Good design is thorough down to the last detail
• Good design is environmentally friendly
• Good design is as little design as possible

What I find most useful about that list isn’t any single line, it’s that none of them mention a tool, a medium, or a piece of software. They’re all judgment calls. That’s the durable layer I keep coming back to, and it’s the part of design education I think we’ve under-invested in relative to how much time we spend teaching whatever application happens to be popular this year.

From maker to director

A lot of design education has historically been about teaching people to produce artifacts: typesetting, drafting, Photoshop, Figma, code. AI is abstracting away a huge amount of that mechanical production, and that shifts where professional value actually sits.

‍

The designer starts to look more like a combination of editor, curator, critic, creative director, and technologist. AI can generate hundreds or thousands of possibilities. The designer decides which ones are worth pursuing. Vision has to come before generation, because without a point of view, infinite generation just produces infinite mediocrity.

‍

The process starts to feel almost improvisational: vision, generate, evaluate, respond, redirect, refine, realize.

‍

The scarce resource becomes judgment. And as AI itself gets better at evaluation and curation, judgment alone probably won’t be enough either. The deeper advantage becomes some combination of vision, taste, context, direction, and imagination, the ability to say which space is even worth exploring in the first place.

As production becomes abundant, selection becomes the valuable skill.

None of this means tools stop mattering. AI is just another tool, potentially a far more powerful one. The future designer needs to orchestrate an evolving ecosystem of tools rather than master a single application.

Ten competencies, not courses

Here’s where I’d actually rebuild the curriculum: stop designing it around courses, and design it around competencies instead. Courses are containers. Competencies are what you actually want someone to walk away able to do.

‍

The question becomes: what must someone be capable of doing to think and operate like a designer, regardless of medium?

1. Communication and articulation. Language, visuals, storytelling, persuasion, and now instructions to machines. A designer who can’t articulate an idea can’t direct people or AI.
2. Critical thinking and judgment. Questioning assumptions, weighing tradeoffs, defending decisions. This gets more important, not less, as producing possibilities becomes nearly free.
3. Problem framing and solving. Understanding ambiguity, finding the real problem, generating and testing possibilities. Design thinking belongs here, as one method among others, not as dogma.
4. Human understanding. Psychology, behavior, culture, accessibility, research. Designers need to understand the people they’re designing for.
5. Form and composition. Typography, hierarchy, color, space, motion. Teach the principles of form, not loyalty to a particular medium.
6. Technology and material literacy. Understanding computation, data, AI, and how contemporary experiences are actually built. Not “learn to code,” but “learn enough to design intelligently with the material.”
7. Making and iteration. Sketch, prototype, build, test, break, revise. AI accelerates this loop. It doesn’t remove it.
8. Systems and context. Products live inside economic, organizational, and societal systems. Designers should be able to move between detail and system.
9. Collaboration. With other designers, engineers, researchers, clients, and increasingly with AI agents. This shouldn’t just be taught, it should be required by the environment itself.
10. Design history and criticism. Knowing what’s been tried, what worked, what failed, and why. Bauhaus, Swiss modernism, the history of human-computer interaction, the evolution of a discipline’s own arguments with itself. Without this, students mistake reinvention for innovation, and repeat mistakes the field already has language for. History is also where judgment gets its reference points. You can’t develop taste in a vacuum.

Let projects be the classroom

Instead of learn, learn, learn, then apply, flip it: encounter a real problem, learn what you need to solve it, apply, get critiqued, iterate, reflect.

‍

A single project might require psychology, typography, AI, economics, writing, and prototyping all at once, which is a lot closer to how design actually gets practiced. And the projects themselves should get progressively more real over time: simulated briefs and AI stakeholders early on, then multidisciplinary work with external briefs, then actual clients and apprenticeships. The constraints should get realer as the student gets more capable.

A three-phase model instead of four rigid years

Rather than a fixed four-year program, I’d think about this as a two to three year adaptive formation period, with students moving through at different speeds. In this new digital era, some of the foundational coursework could be done remotely, self-paced or in an immersive, intensive setting.

‍

Foundation, roughly a year, focused heavily on the ten competencies above. Tool instruction stays deliberately subordinate. Students might start with sketching, writing, physical making, and critique before leaning hard on sophisticated software. The goal is simply: learn to see, think, communicate, judge, and make.

‍

Exploration, where students start building their own combination of capabilities instead of being forced into the same shape as everyone else.

‍

Design + AI
Design + motion design
Design + psychology, business, motion, architecture, fashion, data, whatever fits.

‍

This produces differentiated designers instead of standardized graduates.

‍

I care about this one specifically because I’ve taught the alternative. A curriculum that left very room for electives and requirements that included courses focused on print-based compositions or industrial design.

‍

Forcing every student through that exact frame doesn’t translate cleanly to mobile design, social content, or motion. There’s something underneath that does carry over, but a curriculum built as one fixed sequence of required courses can’t easily separate the durable lesson from the medium it happened to be taught in.

‍

I still believe in the T-shaped designer, broad enough to collaborate across a range of disciplines, deep enough in one or two to actually lead. I believe every student should be given agency to define the shape of their T. Let them choose what goes into the vertical stroke, be it engineering, psychology, fashion, business, whatever pulls at them, and they end up differentiated instead of interchangeable, with real strengths instead of a generic competence.

‍

Lastly, the one final phase of the education would be Apprenticeship, where learning moves into real professional practice. Real projects, real clients, real deadlines, real consequences. Done in partnership with industry, this phase becomes the bridge into employment itself, not a simulation of a job but the job, where practical knowledge finally catches up to whatever theory came before it, and where the relationships built often turn directly into the first offer.

Three speeds of knowledge

Not everything in the curriculum should update at the same rate. I’d split it into three layers.

‍

A slow layer, foundational stuff like critical thinking, psychology, ethics, and systems thinking, reviewed maybe every few years. A medium layer, practice-level things like research methods and interaction paradigms, reviewed annually. And a fast layer, the actual technology, AI models, software, workflows that adapts thanks to apprenticeships and real world practice.

‍

That’s how a school stays capable of changing at roughly the speed of the practice it’s supposed to prepare people for.

The point of all of this

Once everyone can generate a competent artifact, artifact production stops being the differentiator. The differentiation moves upstream, to the questions underneath the artifact: what should we make, who is it for, what problem are we actually solving, which of these thousand possibilities is worth pursuing, and how do we make it exceptional.

AI doesn’t make design education less necessary. It makes the right kind of design education more necessary.

That’s the whole thesis, and it points to a concise mission for the entire idea: Don’t teach students to use the tools of today. Teach them to think, see, judge, create, and adapt like designers, so they’re ready for whatever tools come next.

‍

Enrollment in design, art, and creative education programs is dropping due to widespread industry instability, tech sector layoffs, and anxiety over generative artificial intelligence. Fewer people are entering design at all because they assume AI can just do it. I think that assumption badly misreads where designers are actually needed. The parts of the job that made someone a designer in the first place, vision, taste, judgment, direction, were never the part AI was going to make obsolete. If anything, they’re the part we’ve been undervaluing all along.

‍

This is an early draft of a set of ideas I’m still working through, not a finished framework. It’s grown out of my own experience in the classroom as much as any theory. I’d genuinely like to hear where you agree, where you push back, and what I’m missing.