AI assistants generate working code at frightening speed. The risk isn't that they're often wrong — it's that they're often “right enough” that you stop noticing when they're not.
That gap — between “I shipped it” and “I understand what I shipped” — is where careers quietly stagnate. You feel productive. You close tickets faster. But you're not building the judgment that catches the bug in the next AI-generated PR — the one that ships SQL injection because the variable names looked clean.
A senior developer I watched recently put it bluntly: junior engineers should be rigorously trained before AI becomes part of their daily work. Not as gatekeeping — as protection. AI compounds knowledge. It doesn't fill in foundations you never built.
This site is a self-check. Not a credential, not a real certification. Ten questions across the topics where AI does most of the heavy lifting for you — security, concurrency, testing, performance, architecture, AI-specific traps. The point isn't only whether you can catch what AI gets wrong; it's whether the knowledge underneath is still yours.
If it is: you can keep AI in check. If it isn't yet: you know which foundations to revisit before the gap quietly widens.
It's not gatekeeping. It's a mirror.
— Anton