Owen Gregorian argues that software engineering's importance is durable because demand for software perpetually outstrips supply. AI advances in code generation do not address the hardest parts of the discipline: understanding user needs, creative system design, and navigating complexity. These human-centric skills — empathy, creativity, systems thinking — remain the bottleneck, ensuring that skilled engineers remain essential even as implementation becomes increasingly automated. The key insight is that software engineering's value was never primarily about typing code. The hardest and most valuable part has always been understanding what users actually need (not what they say they need), translating ambiguous requirements into precise system designs, and managing the emergent complexity of interconnected systems. AI can generate code from specifications, but it cannot generate the specifications themselves — that requires human judgment, domain knowledge, and the ability to navigate social and organizational complexity. This positions software engineering as one of the most AI-resistant professions precisely because its core difficulty lies in the human-interface layer, not the implementation layer. As AI handles more implementation, the relative value of the human-centric skills increases. ## Key Principles - Demand for software perpetually outstrips supply, ensuring engineering remains important - The hardest part of engineering is understanding user needs, not writing code - Creative system design and complexity navigation require human skills AI cannot replace - AI automates implementation but amplifies the value of human-centric engineering skills ## Cross-Domain Connections - [[Judgment as Durable Moat]] — Human judgment as the irreplaceable evaluation layer - [[Builder Consumer Distinction in AI Hype]] — Understanding which engineering challenges AI actually addresses - [[Developer as Builder Not Orchestrator]] — The case for preserving hands-on building skills - [[Thinking vs Understanding Distinction]] — Understanding user needs cannot be outsourced to AI