Original article was published by Tarron Lane on Artificial Intelligence on Medium
I Swear it’s the Algorithm’s Fault that Lizzo Shows Up in my Feed.
Algorithms are the unseen third parties listening in to my digital life and trying to exploit my patterns. They just aren’t super good at it… yet.
Allie the Algorithm Fairy
I recently got an email from Spotify, asking if I wanted to try out their new “Duo” package. For a small price increase, the package would let me have two different people create two separate profiles on the same Spotify account. Then the algorithms and recommendations will be personalized for each user on the account, rather than being the abominable mishmash created by a couple sharing an account with two radically different musical preferences.
Kami and I both use the same Spotify account. We have a good chunk of musical overlap, but our musical tastes reach to the extremes on opposite ends on many musical spectrums. The poor subset of Spotify’s recommendation engine that is assigned to our account is probably pulling out her hair trying to understand our musical preferences, falsely assuming us to be a single entity.
I’d be more culturally inclined to personify an intrusive algorithms as a sinister spy cataloging my every move from the safety of its unseen shadowy corner. In contrast, I feel like a more accurate image would be to just picture that out there somewhere in a digital cubical is an overworked office pencil pusher for Spotify Recommendation Co. getting frustrated at every turn while trying to piece together a coherent “Recommended for You” playlist for me in time for the Monday release. I’ll call her Allie. Allie the Algorithm Fairy has a tough and thankless job. Almost daily Allie sees my Spotify account going full-steam ahead on something like Ariana Grande’s “Dangerous Woman” album, before suddenly jump to something as jarringly different as Brooks and Dunn’s Boot Scootin’ Boogie. There isn’t any way for a sane mind to bridge that gap and boil that down to a single recommended list of possible new music to enjoy. Perhaps this email product offering from Spotify for their Duo upgrade is Allie’s way of reaching out in a desperate last-ditch attempt to make sense of the chaotic mess that is my Spotify listening history. “This CAN’T be just one person!” She says, on the verge of tears.
Then, just when she thinks she’s finally gotten a hold on things, the Wiggles’ Fruit Salad song comes rolling down the “Recently Listened To” pipeline and she crumples up her draft recommendation playlist and angrily throws it in the trash to start over. Not to mention the angst she must feel on the occasions when I barely touch the lovingly crafted Christian rock compilation she pieced together into the Monday recommendation playlist, which she based off all the Gospel music I listened to the day before on Sunday.
Artificial Intelligence just joined the chat
More and more, artificial intelligence (A.I.) and algorithms are creeping into all the different parts of my digitally connected life. They’re picking the music I listen to, the route I take while driving, and the next TikTok video I see in my feed. Although the Spotify Duo scenario is mostly just an amusing and harmless consequence of that increasing A.I. interaction, the occurrence hints at the interesting future I will have with A.I.
The idea of an overworked corporate fairy pulling strings behind the scenes is not so far from reality. Algorithms are becoming a third entity that I have to take into account when I interact on the digital scene. On YouTube, I have to think twice about looking up and watching that unique Halloween costume video; it might be the gateway video into the world of cosplay and it’ll be a week before my YouTube recommendations settle back down to normalcy. One click — one algorithmic assumption — and then a week to pay the price.
Is it worth that $4 extra a month for Spotify Duo? Certainly it might keep Ariana Grande and T-Swift off my Spotify home page and save me from coworkers’ judgments when they catch a glimpse during my screen share. Not to mention it might also keep contamination levels lower on my weekly “Recommended for You” playlist. The unease about the offer comes because I’ve never previously had to so blatantly financially quantify my dislike for certain music. It’s an uncomfortable butting of heads between my fragile masculinity and my compulsive penny pinching. Frugality prevails, however, and with $4 on the line I’m inclined to proclaim “Real men wear pink? No, real men aren’t embarrassed to jam out to an unexpected Demi Lovato tune while road-tripping with the boys!”
I wish I could say the blame for all the Algorithmic absurdity within my Spotify account could be placed at the feet of its dual-usership. Unfortunately, I would probably end up in the same pickle even if I was on my own.
For example, I’ve never seen such a scary and disturbing “Recommended for You” playlist as when I searched and listened to a couple catchy tracks from the Muppets 2 Movie over one weekend. It starkly diverged from my normal music patterns, and also seemed to have given Allie the confidence to go-ahead and open up the world of children’s soundtrack jingles and sprinkle them into unexpected places. I can just imagine the bumbling explanations I give to my peers when I get put in charge of the music and unintentionally bring the Smurfs theme song to the party.
Sometimes I wish I could put the Algorithm on pause for a short period at a time. What if I could just tell YouTube to calm down and stop taking notes for a second? I just want to quickly look up that random video a coworker mentioned, but I don’t want it to corrupt my carefully-curated YouTube recommendation feed. I’ve worked so hard on that feed, and finally gotten it to be useful. It’d be a shame to knock it off balance by throwing an unexpected “Too Many Cooks” video into the pattern-crunching recommendation engine. It’ll be like when I searched for step ladders on Amazon and now I’ve seen more commercials for different brands of step ladders than I ever thought could exist. Nobody wants that to happen for the odd things my coworkers tell me to look up.
The Digital Elephant in the Room
I can’t imagine a future where this kind of interaction with recommendation algorithms is not dramatically expanded and exploited by marketing teams and product designers. Everything from video games to self-driving cars are soon going to be watching me and making guesses about what I might be interested in buying, listening to, or visiting. Did I stop by Taco Bell during a particularly low point last week? Now during this week my car will ask if I want to revisit Taco Bell when I ask it what restaurants are open in the area. Great! Right in front of the colleague I was trying to impress!
I don’t drag this reality out into the spotlight in order to declare doom and gloom for society; I’m not yet convinced that the increased interfacing of my daily choices with always-observing algorithms will be the end of the world. Although when I phrase it that way it does send a few shivers down my spine. The other side of the coin, however, is worth the risk in my opinion. I can geek out about the possibilities for video games and virtual realities which will be able to learn about what kinds of features I enjoy and then adapt to give me more of those experiences. A.I. is just a new and powerful tool, and I’m convinced it’s going to be used increasingly in the future regardless of any effort to try to stop it. I only hope innovators come up with sufficient positive applications of the technology to make up for the negative outcomes (such as the invasion of our privacy).
Regardless of what comes next, I am undeniably already getting some good practice spotting Allie as a judging and reactive observer in my life. For now, it’s mostly just purchasing history and musical tastes. Tomorrow it might be my eating habits and driving preferences. Whatever expansion of that interaction happens in the future, I’ll make sure to tip my hat to Allie the Algorithm Fairy each time I pass by.