As I say in this post, the main goal of its early implementation is to get feedback to train the algorithm on a bigger training set. I thought I made it overly clear in this post. Maybe I can add a few caution words on the popup to remind that to users who may not have read this post.