I'll be quite impressed if ML can give meaningful feedback for the subtleties of what is considered flattering.