Blame Spotify. If you are one of its 355 million users, then you will likely know about the end-of-year “Wrapped” roundup. Each December, a jaunty video tells you which songs and artists you listened to most in the preceding 12 months. There’s no hiding from the data. In your mind, you were listening to trap and experimental techno. In reality, you were listening to adult standards and help-me-sleep whale sounds.  Still, if you want to advertise to the world that you are in the global top 1% of Phish fans, this list is your chance. 

It’s a neat idea, for Spotify. The problem is that anything that works in tech is immediately adopted by other companies. And what works for music does not work for every customer relationship. I like to order takeout as much as the next man, but do I need a summary reminding me that I’ve had 30 curries and 26 chicken burgers over the course of our last trip around the sun—plus, just once, ice cream in the morning? Or my banking app telling me how much I spent on “leisure”? Or Uber letting me know that I should really just divert my earnings directly to them, rather than going through the charade of holding the money in my account?¹ Trainline, a UK-based travel booking site, was widely (and rightly) mocked when its 2020 roundup reminded users that it had been a tricky year for long-distance travel.

Where does it end, all this algorithmic nonsense? Can we look forward to cutesy notes from our healthcare providers telling us how much weight we’ve put on? Warnings from the municipality about the number of wine bottles in our recycling bin? These firms want to create an illusion of friendliness, a way to suggest to us that we are not merely another drop in an ocean of anonymous users. But tech firms succeed because they offer convenience. By design, they let you forget you are ordering takeout and getting cabs in the hope that you will do so more often. Nobody wants to stare their own human weakness in the face. There’s enough of that around during the holidays anyway. And there’s a fine line between a charming new algorithmic data tool and that most old-fashioned “end-of-year wrap-up”: the credit card bill.

GOOD IDEA:
HONEST AI

Words by Harriet Fitch Little

Spotify’s Year Wrapped feature launched earnest copycats, but also jokey parodies. Developers Mike Lacher and Matt Daniels at The Pudding went so far as to create a bot that mimics the sites’ process and formula to mock users’ musical proclivities. How Bad Is Your Spotify bases its algorithmic understanding of “good” music on the opinions of dogmatic sites such as Pitchfork, and judges a user’s Spotify listens according to how far they deviate from it. It spits out creative insults such as “learns-about-rap-from-TikTok-bad” or “manic-pixie-dream-girl-bad” before slapping a final rating on your listening habits. The app is ultimately not as personalized as it appears to be, and the AI is not as sophisticated: It gives all users the same prompts and dishes out similar insults with the artists’ names swapped in and out. But this is exactly the same mechanism  Spotify uses to make users feel special. If a bot can be your cheerleader, it can be your cyberbully too.