Why do we test with code challenges anymore?

I haven't interviewed for a job in a while, and certainly not since AI changed the way we write software. So it felt weird preparing the way I used to. Going on LeetCode and practicing code challenges just doesn't seem relevant anymore.

Still, I wanted to be ready for the inevitable live coding portion. I completed several challenges, only to find out the more complex ones are behind a subscription. So I did what any AI-loving software engineer would do: I asked Claude to build a LeetCode-like app I could run locally. Within ten minutes, I was up and running with my own version of LeetCode.

The whole exercise took me from "Why do we still test like this?" to "How do I even have a job if AI can build something like this that fast?"

Anyway, you can find the repo on my GitHub: code-arena. It has some big security holes and it's meant for a single-user, but that's nothing to worry about since it's just a local app. It's not meant to be deployed on a server open to the general public, so don't be stupid with it. Otherwise, it's very usable. You can add new challenges easily, and while it only supports Java and Python right now, adding more languages is easy too.

But seriously, why do we even have jobs anymore?