One of my ongoing projects is to clean up the older parts of this blog (fixing broken links, etc.). That’s how I came across this, written 18 years ago:

So here’s what I want, what I really really want: a computer program that will let you feed in the name of a song and the name of an artist, process for a few seconds, then spit out a believable cover version. For instance, I would like to hear Tom Waits doing “Joy to the World” (the Three Dog Night song, not the Christmas carol — can’t you just hear him croaking out “Jeremiah was a bullfrog”?), or Frank Sinatra singing the Velvet Underground’s “Sunday Morning.” I would pay real money to hear the Sisters of Mercy covering “You’re So Vain” or the Pixies’ version of “I’m So Tired.” In fact, I would probably have the Pixies do the whole White Album, just to hear what it sounded like.

(Here’s a link to the original post. Some of you reading this shared ideas back then.)

With the advent of AI, this is now most likely possible. And yet I hesitate to go there, sensing the presence of an infinite rabbit hole from which I might never emerge.

Your thoughts?