For a long time, production was one of the clearest signals of design skill. A strong designer could move from a vague brief to a coherent flow, explore multiple directions, create detailed screens, build a prototype, and communicate the result with precision. Doing that well required time, practice, and a broad set of craft skills, and I think those skills still matter a great deal. What is changing is the economics of making.
AI can produce plausible copy in seconds, generate visual directions, synthesize research, write interface code, create variants, and help turn an idea into a working prototype far faster than before. As a result, the cost of producing an option is falling, sometimes dramatically. At first, this looks like a straightforward productivity gain, and in many ways it is. But in my view, it also makes something more visible that has always been true: producing more options does not necessarily lead to better decisions. When making becomes cheap, choosing becomes more important.
The bottleneck moves upstream
Imagine a team that can generate ten interface concepts in the time it previously took to create two. The obvious response is to take advantage of that speed and generate even more, but I do not think volume is where most of the value will come from. The harder work is still deciding what problem is worth solving, who it matters to, what the product should optimize for, which constraints are actually meaningful, and what tradeoffs the team is willing to make. These questions are related, and answering one often changes how the others should be understood. AI can help explore this space. It can surface alternatives, summarize evidence, challenge assumptions, and make tradeoffs more explicit. But I think the responsibility for deciding what matters still sits with the people building the product. As production accelerates, the bottleneck moves toward framing, prioritization, evaluation, and direction. That is where judgment becomes increasingly important.
Judgment is more than taste
Design judgment is sometimes described as an intuitive sense of what “looks right.” Taste is part of it, but I think the more important form of judgment is broader than visual preference. To me, judgment is the ability to make a decision when the evidence is incomplete, several outcomes are possible, and there is no obvious correct answer. It requires connecting context, intent, tradeoffs, decisions, and likely consequences rather than evaluating each design choice in isolation. A polished interface may still be solving the wrong problem. A clever interaction may create unnecessary operational complexity, and a feature users explicitly request may still distract from the reason they use the product in the first place. Good judgment means being able to notice those relationships early, before they become expensive to change. I think this ability becomes more valuable, not less, when execution gets easier.
Plausibility is getting cheaper too
One of the more interesting effects of generative tools, in my opinion, is that weak ideas can now become convincing very quickly. A concept can have polished copy, realistic data, smooth transitions, and a well-produced presentation before the underlying logic has been seriously tested. In the past, roughness often gave teams a useful signal that something was still exploratory. A sketch looked like a sketch, and a rough prototype made its uncertainty visible. Those signals are becoming weaker. I think this creates a new problem for design teams: fidelity can now increase much faster than certainty. A prototype may look finished long before the thinking behind it is finished, and that can make an idea feel more validated than it actually is. Because of that, critique becomes more important. Teams need to get better at separating production quality from decision quality and at evaluating whether an idea is coherent, useful, differentiated, feasible, and appropriate for the context, rather than simply reacting to how convincing the artifact looks. A beautiful answer can still be the wrong answer, and AI makes it much easier to produce beautiful answers.
Saying no is part of design
Faster production also makes it easier to add things to a product. Creating another version, feature, state, experiment, or piece of generated content may now take very little time, which can make addition feel almost free. I think this is misleading. The cost of producing a feature may decrease, but the cost of maintaining, explaining, supporting, and integrating it into the rest of the product does not disappear. This is why restraint becomes more strategic. Good products are shaped as much by what they exclude as by what they contain. A coherent product has boundaries, and those boundaries communicate what the product is for, what it is not for, and what the team has decided not to optimize. AI increases the number of things a team can reasonably build. In my view, design judgment determines which of those possibilities deserve to become part of the product at all. That makes saying “no” a creative act rather than simply a limitation.
Seniority becomes visible in the reasoning
Two designers may increasingly be able to produce similarly sophisticated prototypes even if they have very different levels of experience. As tools reduce some of the production gap, I think the difference between junior and senior work will become less obvious on the surface of the artifact.
It will appear more clearly in the reasoning behind the work. An experienced designer is more likely to question whether the problem itself is framed correctly, recognize which constraints are real and which are inherited assumptions, understand how a local decision affects the wider system, and explain why one direction was chosen over another. This does not mean that seniority is simply the ability to justify a decision after the fact. In my opinion, the deeper difference is in the quality of the mental model that produces the decision in the first place. Senior designers often notice different variables, ask different questions, and recognize consequences earlier. AI may help more people create polished artifacts, but it does not automatically give everyone the same understanding of the product, the organization, the users, or the system around it. The artifact remains important, but increasingly it becomes evidence of a decision rather than the decision itself.
Leadership needs better filters, not more output
I think the same shift applies to design leadership. If a team can produce far more concepts, prototypes, and experiments than before, a leader who responds by simply asking for more production is likely to create noise rather than progress. The harder leadership problem is deciding how that increased capacity should be directed. Strong teams need a shared understanding of what good means for their product, which principles should guide decisions, where broad exploration is useful, where consistency matters, and which risks are worth taking. They also need clarity about which outcomes matter enough to influence tradeoffs. In my view, this is one of the most important roles of design leadership in an AI-enabled team: creating useful filters and strategy rather than becoming the person who reviews every generated possibility. A strong design culture allows people to make better decisions without requiring a leader to approve every detail. I think of that as judgment operating at an organizational scale.
The value of design is moving, not disappearing
AI is often framed as a question of whether design work will become less valuable. I think that framing is too simple. What seems more likely to me is that the distribution of value inside design is changing. Some forms of production will become faster, cheaper, and more accessible, while abilities such as framing, synthesis, systems thinking, editing, critique, product understanding, collaboration, and judgment become easier to distinguish from pure execution. Craft does not disappear in this model. If anything, genuinely exceptional craft may become more visible once competent production becomes common. But it will also become harder to confuse polish with purpose. A designer who can make quickly but judge poorly can now produce more wrong things, at higher fidelity, in less time. That is not necessarily progress. A designer with strong judgment, on the other hand, can use faster production to explore more intelligently, test assumptions earlier, and converge with much greater leverage. That is why I do not think the most interesting question is whether AI makes designers faster. It clearly does. The more important question is what designers choose to do with that speed. When making gets cheap, the scarce resource is no longer the ability to produce another option. It is the ability to understand which options are worth pursuing, which ones should be rejected, and why.
The work changed before the role did
There is another shift happening: designers are changing faster than the organizations around them. A designer may now move far beyond a Figma file — writing specs, building prototypes in code, fixing accessibility issues, working directly with components, and contributing to production-ready implementation. But many companies still evaluate that work through old categories like handoff, design artifacts, and traditional role boundaries. In other words, people have upgraded, but the operating system of the company has not.
I think this creates a growing mismatch between the value designers create and the value organizations are able to see. Roles, performance reviews, and hiring models need to catch up with how the work is actually being done.
I’ll explore what this means for roles, performance systems, and AI design tools in future articles.
