AI Makes Coding Easy, But Deciding What to Build Remains the Hard Part

Decades of developer tools — from Stack Overflow to GitHub to AI coding assistants — have steadily reduced the effort required to turn an idea into working software. However, while AI can now generate prototypes in a weekend, it cannot answer the more fundamental question of whether something is worth building in the first place. The author describes personally spending a weekend building a fully functional app with AI assistance, only to realize afterward that no one — including themselves — actually needed it. As implementation costs fall, the risk of shipping unnecessary or unmaintained software grows, since the traditional friction that filtered out bad ideas has largely disappeared. Experts argue that skills like problem framing, user understanding, and knowing when not to build are becoming more valuable precisely because the act of building itself has become so cheap.