Can AI avoid being affected by insulting commentary?