If you pick really small and simple little ideas, you can finish them quickly and I think that way it is easier to pick an idea for the next project.
Personally I think games are a good target if you want to explore a lot of things, but I think it is important to keep things practical and get going / actually work on it, instead of just accumulating more and more ideas.
Because that practical experience allows you to gauge which ideas you could quickly realize and which ones are manageable vs too big.
Well you always could start with something more like GTA1/GTA2 ;), or if you want something simpler you could create a Bulletheaven/Vampire Survior type of game (or anything else you find interesting).
Then break that down into even simpler parts:
- open a window and have something rendered
- add some keyboard controls to change what is rendered
- move the camera
- have some player entity
- make it shoot a bullet
- add an enemy
- make bullets hurt entities
- pick up an item
- …
I think over time you will naturally have more and more ambitious ideas for what you want to add / improve about the game and that will lead you down specific topics like how to do spatial partitioning (e.g. via a grid) so you can avoid slow collision checks etc.
I might be biased, but my short answer is start with raylib and get stuff on the screen, get it to move, load and play sounds, then try to find something that is fun and iterate on that. Raylib example using the package manager