Curating is becoming a buzzword, applied not just to art but to your Instagram feed, your Spotify playlists, your wardrobe, the contents of your fridge… Curating has been democratised: no longer the preserve of elite tastemakers, the internet’s capacity for sharing content has enabled us all to be curators. But what does curating really mean? Gathering together carefully selected items is the simple part we can all do, but providing a platform to showcase and champion them is just as important. And gaining trust so people believe it’s worth their time to try something new. Not to mention deciding how best to present things in time and space for maximum impact, and creating a journey of discovery for audiences. As someone who programmes a festival, I suppose I am a curator of music and sound.
Curators are a product of human experience, and every artistic choice is influenced by knowledge or experience gathered. But what if the curator is not a human? What if it is a computer algorithm?
Back in April I attended a talk on ‘the role of human curation in the age of AI’ as part of the conference at Tallinn Music Week in Estonia. The (entirely human) panel were debating whether algorithms used by music streaming services can replace human curators, i.e. DJs, radio producers, journalists. All music streaming services use artificial intelligence (AI) to recommend other music you might like, to create ‘radio stations’, and to create playlists. With the massive volume of music now available online these algorithms are vital to help you find your way; we’ve all experienced that ‘search bar paralysis’ when we don’t quite know what we want to listen to.
Previously the selections offered by these algorithms have been a little limited, and in fact I learnt at the conference that Spotify’s Discover Weekly playlist only began to take off when they combined choices from actual listeners with the computer-generated recommendations*. The tech is improving at a fast pace, but often an algorithm still takes you down a steady path of similar music without the ‘wild card’ selections you might get from a DJ (although I’m sure wild cards are on the horizon for algorithms). As far as I know, no-one has applied AI in the context of live music yet, but it wouldn’t take much work to feed previous festival bookings in to a database of artists and generate a new line-up.
Hans Ulrich Obrist (Artistic Director of the Serpentine Gallery) has a book on curating where he describes his role as that of a ‘junction maker’. A junction is a place where things/objects/sounds/people meet each other, and he stands at the intersection. Often this approach creates surprising connections between people, across decades or styles. For me, this is the one thing human curation can do that AI has not cracked yet. To take another example from the art world, I recently visited an exhibition at Tate Liverpool which puts Tracey Emin’s work alongside Romantic poet and painter William Blake: I can’t see how a computer would ever make that connection. As human curators we can provide deeper contextual links, beyond just music of the same genre or things that sound similar. Those links might be anything from current affairs to how intense something sounds, or how it makes you feel. Until a computer can encapsulate a lifetime of experience and feel emotion, I think my job is safe…
*Update: if you’d like to find out how Spotify’s Discover Weekly works in more detail, check out this interesting piece. (Caution: contains maths.)