Let me explain my perspective as I do vibe coding for some side projects. AI (and even if it works correctly) is a thing filling the blanks. Still, the value depends on how much work you put in.

For recent ones, it is a interactive visualization of StarCraft 2 (https://github.com/stared/sc2-balance-timeline). Here I could do it myself (and spend way more time than I want to admit on refactoring, so code looks OK-ish), but unlikely I would have enough time to do so. I had the very idea a few years ago, but it was just too much work for a side project. Now I did it - my focus was high-level on WHAT I want to do and constant feedback on how it looks, tweaking it a lot.

Another recent is "a project for one" of a Doom WAD launcher (https://github.com/stared/rusted-doom-launcher). Here I wouldn't be able to do it, as I am not nearly as proficient in Rust, Tauri, WADs, etc. But I wanted to create a tool that makes it easy to to launch custom Doom maps with ease of installing a game on Steam.

In both cases the pattern is the same - I care more on the result itself that its inner workings (OK, for viz I DO care). Yes, it takes away a lot of experience of coding oneself. But it is not something entirely different - people have had the same "why use a framework instead of writing it yourself", "why use Python when you could have used C++", "why visiting StackOverflow when you could have spend 2 days finding solution yourself".

With side projects it is OUR focus on what we value. For someone it is writing low-level machine code by hand, even it it won't be that useful. For some other, making cute visual. For someone else, having an MVP that "just works" to test a business idea.