I recently saw the last image in an article I watch was tagged as having no fair use rationale. I decided to check it out, and saw this on the image tag: "Administrators: check the image talk page for comments before deleting the image."
I don't know a lot about the whole fair use policy, and I don't really care to. Image policy in general is absurdly convoluted at times (mostly due to copyright laws, I imagine). I simply put a message on the talk page, mentioning why I thought fair use might apply, so that the administrator could take a look (http://en.wikipedia.org/wiki/Image_talk:ChagallGuevara.jpg). If I was right, I figured it would be worth going back to try and find the right template and sort it out, but I didn't want to waste the time if the image was just going to be deleted anyway.
Instead, the admin who deleted it never bothered checking the talk page at all. It was one of hundreds of images deleted in minutes using Twinkle. There's no possible way he even checked if the talk page existed at that deletion rate. Functionally, it might as well have been a bot doing the deleting.
Now, I really don't particularly care about images, for the most part. What does bother me is that one thing was said, but something completely different was done. This is a very unfriendly way to do things. If we're going to automate the entire process of tagging and deleting images, it needs to be clear, along with a way for people to ask for review by other editors, much like PRODding an article.