I'd love to know how the individual curated playlists are done, as I suspect that they have data on what songs people who like one song also like and can use some form of AI to create playlists by compatibility and even song-to-song fit.
A week ago Tuesday, my wife and I took our daughter and grandbaby to San Francisco and spent the day in the city before dropping them at SFO for their flight home to Georgia.
On the drive back to our house (about 2 1/2 hours because we left SFO around 6:00 p.m.), I decided to try "Michael Hagerty's Station" for the first time. Apple put it together based on what I've played over the years and attempted to extrapolate my tastes.
Since it's a "station", you can't look at a list. So I just hit "play" and figured we'd see how well Apple knows me.
It started off very strong for the first half-hour, a pretty much flawless mix of 50s, 60s and 70s jazz and R&B groups like Average White Band and Tower of Power, with excellent flow between tracks. After that, though, it tried to fold in other genres I like and kinda lost the thread.
To be fair, though, my tastes are
very eclectic. Back in the day when Pandora first hit, I gave that a try. For those who haven't, you give them a few favorite songs and it tries to come up with a "station" that matches your taste. You "thumbs up" the ones you like and "thumbs down" the ones you don't.
I'm sure it was doing exactly what the algorithm said would make the most sense, but it never held past the fifth song.
My joke is that after 20 minutes, Pandora would stop playing music and start sobbing
"What do you WANT from me???"
My daughter, in her teens at the time, said that no matter what she started with, the fifth song would always be Swedish Death Metal.