Is it expected behaviour that Google also crawls #! urls?

I'm a developer for an AJAX webapp and, following Google's specification for crawlable webapps, we use #! to indicate that it's a AJAX application such that we can serve a static page to Google instead. This all works perfectly fine: Google fetches the ?_escaped_fragment_ URLs instead.

However, in the logs we found that even though we follow this specification Google also fetches the original AJAX pages, and in the process it generates script errors.

Is it expected behaviour that Google visits the AJAX URLs, even though it knows we have specially prepared pages for it? I can imagine that Google does this to train it's AJAX crawler, but I cannot find any information about it.

Additionally, does this have any influence on our ranking?