Ask HR: LLMs can't solve basic logic problems, much less code

An LLM can `leet code` because all those snippets were in its training data. It might be able to make smaller variations... but that's it.

If it was such a game changer, we would have seen a bloom of cool KISS apps all over the place. Plus many repos would use LLMs as a LINT.

So let's curb our enthusiasm a bit (unless you show me your fantastic app written in a weekend using vibe coding :P )