Is AI about to expose just how mediocre most developers are?

AI is exposing a harsh truth: most code is crap and many developers are merely mediocre. When AI can generate comparable or slightly better code instantly, the market value of low-quality output collapses, and the industry must reckon with the cost of mediocrity.

The author examined Lovable, an AI that builds entire applications with minimal human input. Its output is not great, but it often matches the worst of human-written code while avoiding some obvious cruft like commented-out sections, stray TODOs, and meaningless variable names. However, the deeper architectural flaws-tight coupling, duplicated logic, and anti-patterns-remain, because the AI is trained on the same flawed codebases humans produce.

Because AI speeds up production, the consequences of crappy code arrive sooner. Start-ups may hit the maintenance wall before an exit, and large enterprises risk critical systems becoming unmaintainable well before budget cycles. Mediocre developers lose the illusion that speed alone grants seniority; their weaknesses become visible in real time.

Good engineers who practice modern testing, refactoring, continuous delivery, and product-focused design become far more valuable. In an AI-assisted world, these practices are no longer optional-they are the only way to extract real benefit from code generation tools.

The industry faces a supply gap: only a small fraction of developers possess deep engineering experience combined with the ability to use AI tools effectively. Leaders must invest in upskilling to avoid a widening talent shortage and the escalating cost of technical debt.