As we live in a historical moment in which the opportunities to create music and art multiply every day, the rapid and constant development of artificial intelligence has become a sheer revolution. Artificial Intelligence has introduced (and keeps on introducing) a number of unconventional approaches to music that were unimaginable just a few years ago, improving (or at least increasing) productivity and making pretty much all musical practices, from composition to post-production, increasingly accessible. The other side of the coin includes several - and legitimate - doubts for many artists. Artistic originality, not to mention years and years of professional education, training and skills in music, risk being lost in a vaguely dystopic future where, with a simple software plug-in, one can create an entire song in few minutes. Conceived within a semiotic framework, the present paper aims at categorize some of the most important and recurrent employments of AI in musical creation and music industry and to propose some reflections on this controversial issue.While, by far, not something that concerns only popular music, AI seems to be particularly relevant for the latter: by consequence, the paper will primarily focus on case studies that belong to the realm of mainstream and alternative pop and rock.