A Reflection on AI, UX, and Design Process
Recently, I had an unexpected moment at work. I logged into the product and noticed that it looked significantly different from what had been designed.
At that point, I had been with the team for a few months, helping take the product from concept to alpha. We had designed and aligned on a substantial set of features, working through brand alignment and high-fidelity designs in collaboration with leadership and development. So seeing a shift in direction without that same shared context was surprising.
From a brand perspective, there had already been healthy debate. Not everyone initially aligned on the selected direction, but we ultimately chose a path that had stakeholder support and fit the broader vision. During feature discussions, AI-generated designs were occasionally used as a way to explore ideas, which I was open to — they can be helpful as a conversation starter or inspiration.
After reviewing the product changes more closely, it became clear that some of the redesigns were driven directly from AI-generated outputs. The stated intention was to prioritize functionality, which is always an important goal. However, when we reviewed the designs together, gaps began to surface — particularly when compared against the existing experience and broader system patterns.
This highlighted something important about product design: good UX work is deeply contextual. It’s informed by desired outcomes, existing platform patterns, user needs, technical constraints, and competitive analysis. AI can generate layouts and ideas, but it doesn’t inherently carry that accumulated context unless it’s intentionally guided through it.
To be clear, there were good ideas in the AI-generated designs — thoughtful layout explorations and even some feature-level inspiration. The challenge wasn’t the tool itself, but the attachment that can form when a design jumps straight to high fidelity. When something looks “finished,” it becomes harder to critique objectively or let go, regardless of how it was created.
This is why I strongly believe in spending time at the conceptual stage. Whether that’s sketches, diagrams, or low-effort exploration, it creates space for discussion and iteration without emotional investment. It’s easy to erase a drawing. It’s much harder to walk away from something that feels complete.
AI is a powerful accelerator, but it isn’t a replacement for UX work. At its best, it’s a source of inspiration or a starting point. Fully relying on it to design products risks producing AI-generated experiences rather than truly user-driven ones.
There’s also a broader team consideration. When design work bypasses the people responsible for it, even unintentionally, it can impact morale. It can signal that the role of UX is optional rather than essential. In many cases, this isn’t about individual decisions — it’s about organizational maturity and trust in the design process.
The takeaway for me has been clear: AI works best when it supports strong design practices, not when it skips them. Tools evolve, but the fundamentals of UX — empathy, context, collaboration, and iteration — still matter.