I think "Open World" is a term that causes a lot of confusion. Building a game of GTAV or Assassin Creed's calibre is NOT in any way achievable by one person. There are thousands of art assets. Hundreds or thousands of animations (depending on animation requirements and how many types creatures that have their own rigs there are), hundreds of sound effects, hundreds of pages of recorded voice acting, dozens of hours of scripted quests, and more. That's not even including the programming requirements.
This doesn't mean making an open world game is impossible for one person (one as deep and as detailed as the AAA titles however would be). If you develop a fairly detailed design document, you can break down all of the assets you need, come up with time estimates it would take you to make the assets, or code the feature, etc., and see roughly how long making the game would take. If it's too long, revisit your design and scale back.
I personally don't think having a "dream game" is a problem. I have lots of dream games that are completely unrealistic for me to build (and I've build AAA games for the last 8 years, so knowing how to do it is not the issue). Sometimes working on my dream game gives me ideas for my other games. The problem is if you let your dream game stop you from working on achievable games, or from learning.