AI but cue the music
Spotify guesses what I’ll like. AI helps me understand why I do.
What happens when we use technology not to automate our choices but to understand them?
I played violin growing up but set it aside in college—and with it, classical music. When my interest finally resurfaced during the pandemic, a few pieces I still loved came with it, including Dvorak’s New World Symphony, Chopin’s Etudes, and John Adams’s Grand Pianola Music. Spotify offered little help finding new pieces, defaulting to Bach, who made my Wrapped list last year even though I never once selected a song by him.
So I set out to see if AI could do what Spotify couldn’t: help me find classical music I loved and do it without making me want to give up on the project.
I decided to start by having AI set up a series of side-by-side comparisons for me, and it gave me eight pairs of songs to listen to–Mahler vs Sibelius, Bach vs Schubert, etc. For each, it named a stylistic difference, e.g., architecture vs emotion.
I cued up the playlist while tidying around the house, pausing to jot notes as I went: the piece I liked more, whether I even really liked either of them, and if something stood out–the simplicity, the dynamic range, the sweep of the violins at a particular timestamp–whatever it was that I could name.
After giving it my feedback on these pieces, AI offered to make a playlist for me based on its understanding of my tastes. It picked ten pieces. I liked all ten, and I loved four, which would be an unprecedented success rate for Spotify. But there was something deeper to it: it felt like AI wasn’t just surfacing broadly popular music from the genre (Bach!) but music that was specific to me.
I looped back to AI the four pieces I loved, and, without prompting, it drew thematic connections between them & other pieces I’d liked:
Natural emotional motion—music that flows like a river, not a staircase. Think Dvořák, Schubert, Brahms.
Sacred stillness with glow—stillness that isn’t empty, but charged. Pärt, Glass, “Silentium,” Rêverie.
Transcendent line and atmosphere—melody that floats, sound that becomes sky. The Lark Ascending, Ravel’s Adagio, Vaughan Williams.
Even though I couldn’t have articulated these themes, I didn’t have to think about whether they felt like me because I instinctively and immediately knew that they did.
That clarity helped me articulate something else: when I collaborate with AI to uncover my taste–whether for music, art, or wine–it’s not about improving some black box algorithm, it’s about becoming a person who can see, name, and shape her own taste. AI isn’t stripping away my humanity: it’s giving me a path to understanding what moves me and why.
Coauthored index: Sarah 80% | AI 20%
