I think there should be something on the community sites themselves to suggest on which community your question belongs.
Because I have a question and i have posted it over 4 communities, without any result.
If you aren't sure where a question belongs, you can ask for a site recommendation here on MSE. Just tag the question with site-recommendation.
We don't have any artificial intelligence for this yet, it seems like a pretty complex project to build. Especially since Stack Exchange sites aren't static, their scope changes over time - usually the scope becomes more narrow. Also, sometimes specific sites split off.
Then again, we have some very creative bot builders in the chatrooms, so maybe some of them would like to pick up the challenge. It's a pretty big AI challenge, that's for sure. Don't expect anyone to build this quickly; it involves NLP and understanding what's being asked in the question.
Keep in mind that some things are off-topic on every Stack Exchange site. For example, polls ("what is your favorite x") are simply not a good fit for the Stack Exchange model.
Think of it this way: If an automated process is able to understand from question body and question title what the OP wants to know (and hence can redirect OP to the best site on the SE network), the system itself becomes obsolete because duplicate flags could be handled automatically, not-an-answer flags could be handled automatically etc.
Furthermore, scopes of site are not something constant and are sometimes not represented by questions asked on the site (sub-aspects are underrepresented and might get ignored by the algorithm, or some questions might just not get closed as off-topic because of a lack of attention).
The lesser amount of off-topic questions asked by users on the wrong sites because of themselves will be mitigated by the amount of off-topic questions asked by users on the wrong sites because of this algorithm.
This will likely lead to Scope-WarsTM (But the System recommended me to post here, why are you closing it then) and, as a result, in confusion about site scopes and even more site-suggestion questions.
Basically the algorithm would just move the problem from unknowing users to faults in the algorithm itself. And it’s always easier to teach users than to constantly fix the algorithm.