Generators definitely have their place. They're good for prototyping, and creating internal tools. Tasks that are time sensitive and can really benefit from eliminating boilerplate code.
Larger LOB, or public facing apps require more thought and planning to ensure that they get the job done right.
You don't have to treat generators as CRUD only things though.
What's wrong with creating an app generator that spits out enough stuff to get you going on a certain app?
If I don't have to spend 15 minutes making a bunch of small changes and additions every time I make a new project then I will be more likely create more projects.
It's part of the reason why building a SOA is painful. If you have 25 services that means you need to start 25 new apps. It might take 20 hours of boilerplate garbage that you have to do just to start developing the business logic.
With generators you can turn that into almost no time at all, you would be bound by how fast your hard drive can write the files to disk or how fast you can pull things from github.
Btw I also use generators for public facing things. Why? Because typing 1 command which generates 6-7 files and also gets me going on what needs to be replaced is a lot faster than starting from absolutely nothing.
It doesn't matter if I end up replacing 100% of the view code or 90% of the controller code. Also if I feel like I will be doing a lot of CRUD'ish things in similar ways then I will spend 15 minutes once up front and create custom generator templates for that specific project.
It's nice to be able to type 1 command out and get a fully working scaffold with your custom theme and changes already applied complete with elastic search integration and faceted navigation, etc.. I do this all the time when applicable.
Larger LOB, or public facing apps require more thought and planning to ensure that they get the job done right.