Introduction

Gen-AI is considered good at visuals. As an MFA student who spends late nights prompting tools for concept exploration, I've seen it generate typography, balanced layouts, and colour palettes faster than we could by hand. For moodboards, quick iterations, and visual brainstorming, it's become part of the toolkit. But brand identity is a different problem than visuals, and that's where the cracks show.

Feed an AI tool a detailed brief, complete with target audience, brand values, and tone, and the output is usually competent. It's also often forgettable. The logo looks like it belongs to a brand, but not the brand you are working on. It lacks the specific memory, the inside joke, the stubborn design choice that comes from a founder pushing back on a designer's first three drafts.

That's the gap. AI doesn't know what it feels like to inherit a family business, or why a certain shade of blue reminds a founder of their first studio apartment. It optimizes for "decent design" as a category, not for the irrational, personal reasons a brand exists in the first place. Logos aren't just visual marks; they're compressed stories. AI can compress visuals. It can't compress lived experience..

I don't think this makes AI useless in design. It's an extraordinary tool for speed, iteration, and exploration. But when I compare AI-generated logos to ones built through actual client conversations, sketching sessions, and revision arguments, the difference is obvious. One has information. The other has intention.

Design school hasn't taught me to fear AI. It's taught me to notice exactly where the human still has to show up.