AI design generation excels at safe outputs but struggles with novelty and constraint¶
Insight: Despite 2.5 years of AI hype, no measurable increase in designer productivity or job displacement has occurred. AI generates safe, average outputs effectively but falters with novelty-requiring work (distinctive logos, high information density, tight constraints). Complex projects show quality degradation as scope increases beyond boilerplate.
Detail: Good design emerges from comprehensive discovery processes, not isolated AI conversations—one-off LLM chats lack context about business goals and brand identity that human designers develop through iterative dialogue. Architectural limitations mean AI struggles outside training data patterns. Viable use cases: personal tools, small public sites, design portfolio projects, supporting content (images/video), and developers building small applications independently.
Sources
- Erik Kennedy — "Where's the AI Design Renaissance" (2026-03-25)