AI Is Dogshit at Design
The article argues that AI tools are fundamentally poor at design because they lack understanding of context, intention, and human nuance, producing generic and often unusable results despite impressive technical execution.
Background
- Tom Creighton is a veteran designer (web/interaction/product) who has been writing about design craft since the early 2000s. The essay is his sharp, experience-grounded critique of using generative AI for design work.
- He argues AI tools (LLMs, image generators) fail at design because they cannot understand context, intent, constraints, or audience — they produce output that looks plausible but is functionally hollow.
- This is a rebuttal to the "AI will replace designers" hype cycle that peaked in 2023–2024. Creighton positions design as a discipline of judgment, empathy, and iterative problem-solving, not pattern-matching or aesthetic output.
- Key distinction: AI can generate "design-y" things (logos, layouts, copy) but cannot *design* — it lacks the ability to frame problems, make trade-offs, or defend decisions based on human needs.
- The piece resonates with working designers who feel the tech industry overstates AI's capabilities while undervaluing their own craft knowledge.