I must admit I don't know Swift yet, but here are my thoughts. Putting some type annotations in here and there really helps produce better type error messages. Return types and arguments are a great checkpoint that'd use for this. Given a type inconsistency in a function application, a full type inference system doesn't know if you messed up the argument or the return value; it just sees that they are incompatible. By annotating, it we can basically tell the user they are (1) returning a value of the wrong type, or (2) taking a parameter of the wrong type. (The same could be said for an inferred assignment, but we want the inference to do some of the work, just not all of it.)
I am getting this from extensive Haskell experience, which performs full inference. Function signatures are generally the smart place to add a type annotations; then let the type system do the rest within the function.
Other benefits:
It improves program comprehension and readability. Seeing the type next to the function helps us understand it quicker. We don't want to have read code and do type inference ourselves to understand what it returns.
Extra type constraints speed up the inference process. I'm not totally convinced anyone cares about this too much. Maybe Apple does in their case.