Lost in Translation
Soft Skills
Design Leadership
The deck was thorough. Research solid, narrative clear, likely questions covered. Then the CPO asked something that wasn't in it, and... silence.
To paraphrase one old military quote: “Preparation built for a controlled environment rarely survives first contact with a live room.”
The case for automating repetitive work is reasonable - if a tool handles the mechanical parts, designers can focus on higher-order thinking. What that argument skips is that repetitive work is how familiarity gets built. Rebuilding the same component type multiple times teaches you the decision behind it. Laying out grid variations by hand makes you intimate with why one works and another doesn't. That intimacy is what lets a designer articulate their reasoning under pressure, defend a decision when it gets challenged, and answer the question that wasn't in the deck. When AI handles the iteration, the designer gets the output without the understanding the process would have produced.
Which means when someone asks why a specific decision was made three screens back, the honest answer is sometimes that the model generated it and it looked right.
Async communication compounds the same problem differently. Writing a Slack brief, dictating a prompt, dropping a Figma comment - none of these train the skill you need in a live conversation with a skeptical stakeholder. Reading when the energy in the room shifts. Knowing when to stop defending a position and start listening. Recognizing that the objection being raised isn't the actual objection. These require real-time practice in real-time conditions.
AI in Design 2026 notes it plainly: AI got excellent at generating possibilities but still struggles with intent, which means that designers who delegate too much of the process stop practicing the conditions where intent is the only currency that works.