@karrisaarinen - Output Isn't Design, Says
@karrisaarinen on X Karri Saarinen argues that design is about understanding the problem and finding a good fit with its context, not about producing polished interfaces quickly. He warns that AI tools that generate outputs fast can hide unresolved problems, leaving products that look impressive but fall apart in use.
- Christopher Alexander defined design as the search for a good fit between form and context, where context means the full set of forces: human needs, technical constraints, habits, and edge cases.
- AI generators produce plausible outputs quickly but often skip the work of shaping the problem, so products look polished yet feel brittle and poorly integrated.
- Saarinen prefers designing visually over prompting because moving things around and testing relationships is how clarity emerges, much like writing clarifies thought.
Karri Saarinen's core argument is that design is not the production of form but the understanding of a problem well enough to know what should exist at all. He pushes back on the current wave of tools that promise to generate interfaces, turn words into products, or collapse design into code. Those tools have their place, he says, but they are not the design process. The hard part is the thinking that comes before any form appears, and generating outputs quickly can hide the fact that the problem was never really understood.
He grounds this in Christopher Alexander's *Notes on the Synthesis of Form*, where design is framed as finding a good fit between a form and its context. For Alexander, context is not backdrop. It is the full set of forces at play: human needs, technical limits, conflicting requirements, habits, edge cases, and relationships that only show up after you spend real time with a problem. Bad design, in this view, is what you get when those forces stay unresolved. Saarinen's concern is that AI-generated work often skips this step and produces polished, impressive surfaces that fall apart in actual use. The form is there, but the fit is not.
Saarinen says he still prefers designing visually over prompting, because moving things around, testing relationships, and refining structure is where clarity comes from. He compares this to writing, where the act of putting words down forces you to organize your thinking, while asking AI to write for you produces text without rearranging your thoughts. AI can still help with prototyping, exploring, and occasionally surprising you, but that is different from design, which needs judgment, conversation, tension, and time. The risk he flags is simple: mistaking generated form for a solved problem.