If the Singer Is a Robot

There are opportunities but also unknowns in the increasingly widespread use of Artificial Intelligence (AI) in the world of music. Then, it also happens that some limitations can turn into opportunities. The recording industry knows very well how major technological shift can upend its business, having already gone through various evolutions, from the age of peer-to-peer filesharing to current streaming platforms. For some time, AI has been creating virtual artists who constantly interact with fans, demonstrating that virtual agents know the fans' names and habits and therefore make them feel closer to their idols. The AI ​​knows how to predict what a user is likely to want to listen to but also knows the exact moment, and the probable why, to interrupt a song to switch to another. "To tell the truth, Artificial Intelligence also composes music and the second AI song contest has already been held, an international competition supported by Eurovision and reserved for musical compositions born from algorithms", says Eleonora Bianchi, head of digital services & consumption at Universal Music Italy.

So robotic creativity is what the future has in store?
Excessive use of automated composition involves risks, such as losing part of the authorship and emotionality associated with listening to a song. From the point of view of production, then, it could reduce the market shares of record companies, with repercussions on employment levels, and race-to-the-bottom effects for artists. A virtual rival can publish a song at a lower cost. However, I believe that the human role will remain.
Why?
For instance, when two rappers for some reason come into conflict with each other, if AI ​​continues to offer them in the same playlist, the two artists and their respective fans may not like that and choose not to use the platform anymore. If anything, I would like to make a better use of new technologies to expand listening opportunities. For example, enriching songs with more metadata. The playlists at the top of the preferences are those for special occasions, which answer questions like "I want music to relax" or those related to a specific event. Given the great impact of the Sanremo festival, the Sanremo tag has gained great relevance. The playlist dedicated to Sanremo songs entered the Top 10 in the world on Spotify and Amazon Music.
What information obtained with AI can, on the other hand, help record production?
Universal Music Italia and other record companies have internal divisions of data scientists, who work on the information provided by streaming platforms to better profile who listens to what. For record labels, these data are useful to draw a whole range of important information both about their artists and to find new singers among emerging or independent ones. We are talking about a form of scouting that, every day, analyzes 200-300,000 names.
And what are you interested in knowing about your artists that you don't know already?
We can predict the average lifetime of a song and decide, accordingly, when to launch a new song by the same artist. We use big data to plan an ongoing calendar of releases in what is a fiercely competitive market.


Eleonora Bianchi graduated from Bocconi in Management. Of her period of study, she remembers above all that "those years were intense, faced with great enthusiasm, thanks to which I was lucky enough to build a network of acquaintances that still accompany me, both from a personal point of view and a professional one. The mindset acquired at Bocconi was crucial to approach the world of work, where I still try to maintain the same enthusiasm and daily desire to learn, which are the bases of success for any Bocconi student and for any manager."