Artificial World or Artificial Intellegence?

In certain pre-development projects, they have a virtual robot instead of a regular robot. Would the habit of creating a modular virtual wold be a habit I'd need to get out of? Because that's been the general approach I've taken with Ruby.

Basically I construct the AI as if it were multiple programs that operate co-dependantly, with them able to be bug tested individually. Then ends up creating more a world environment, and not so much the intellegence in their own right.