Writing about AI and creativity in 2025 sometimes feels like fiddling while Rome burns

Anybody got a Sora invite they’d like to waste on me?

Check it out— two chapters from The AI Music Problem are now Open Access. Download and read for free! www.taylorfrancis.com/books/mono/1...

Last week to for vote for my panel "Can You Hear the Machine? The Disappearing Tells in AI Music" for South By Southwest 2026! Click the little heart in the link below to keep us in the running! participate.sxsw.com/flow/sxsw/sx...

I can use your vote! I'm on a proposed panel for this year's South By Southwest @sxsw.com about music and AI w/ a dizzying lineup of co-panelists: @adamneely.bsky.social, @charlieharding.bsky.social (of @switchedonpop.bsky.social), and @Tristra Yeager (of @RockPaperScissors inc.) 1/2 w/ voting link:

Heavy way to start a morning, @mikeduncan.bsky.social, with the fall of the Girondins… I mean fall of the Mons Cafe crowd

New article out! It's a collaboration in the journal Music Perception where we show that metric emphasis influences what listens hear as tonic/stable in pop music loops: static1.squarespace.com/static/54518...

This article feels like a clear example of techo panic that we'll all make fun of in 10 years. Note the use of #anecdata rather than any actual data in the reporting. #AI www.nytimes.com/2025/01/15/t...

I'm conflicted about this footnote I just wrote: a thread: I would only ever venture to place these kinds of thoughts in a footnote, but LLMs do provide a certain comfort for those of us working in esoteric text-based scholarly fields. ...

Everyone who receives an email from @kevinroose.com this week is going to be wondering whether his #AI wrote it...

Should @theorymeg.bsky.social and I create an early-music-themes #memecoin called #dufaynance? Yes or no or heck yah

Our paper about how we perceive AI-generated music was one of the TOP DOWNLOADED MUSIC PAPERS on SSRN over the last 60 days (with a grand total of 33...) Get my totals up with a download of your own: papers.ssrn.com/sol3/Deliver...

Got a working paper on Music and #AI with a bunch of coauthors. Gives a lay of the land in industry and academia. And, because AI has the potential to change how we think about expression, art, and education, we argue humanistic thinkers need a seat at the policy table. zenodo.org/records/1453...

Another problem with #music #GenAI: it's tough to actually get crisp, reliable datapoints out of audio. Here's a graphic showing what an #AI might "see" when it "listens" to Lizzo's Truth Hurts. OODLES of mistakes. #MusicScience #MusicTheory

Re Musical #GenAI, musical datasets are MUCH smaller than those in other media. Here's an infographic showing when a human would need to start reading/listening/watching to consume the whole dataset. In a landscape of ice-age-length data, music lives in a world of American and Maoist revolutions

OK people, let's take it down a notch and calm down with a gentle sleigh ride. #CursedCarols

The 1981 chipmunks Christmas special is among the most underrated holiday content out there.

As far as I can tell, the Bill Murray Xmas special on Netflix just contracted a dozen talented folks, rented the Carlisle for three hours, turned on the cameras, and taped a bunch of one-take Christmas musical numbers. AND I LOVE IT. Strong recommend.

“Ok hear me out— what if she blows out all the lightbulbs in Emerald City in her last belted sustain?” “But that doesn’t make any sense” “And what if there’s pulsing green lightning in the clouds behind her” To whoever said yes to that ridiculousness: thank you

Reviewing an article with lots of content from #ChatGPT, and English seems to not be the authors' first language. I'm torn about how to think about this. Love the cultural accessibility AI allows, but lots of sentences aren't "by" the authors, and (worse) don't communicate their intended ideas

Musical creativity both recombines inherited materials and adheres to set expectations. Here's a graphic I made: music boxes adhere to expectations but make nothing new, R. Scott's Electronium made new music but 100% randomly. #GenAI and #Humans are special bc they maximize both

Just a friendly reminder you can find “the love is gone” in the “Extras” of muppet Christmas Carol. (Also great ex. of mixture)

1) The ways keys are NOT transpositionally equivalent (e.g., how pieces in B major behave differently that in Bb) and how that changes over time (with @theorymeg.bsky.social) 2) Basically 1% of atonal 20th-c. music actually uses strict transposition, even tho we obsess over it in the classroom

Love seeing all these new people from the #musicNerd #musicTheory and #musicScience worlds on bluesky! Welcome friends!

I had a great time at the University of Delaware discussing music, #AI, and education with the #musicEd students! I also worked with the composition students, showing them how to build a music-making GPT. Yay #Delaware!

If you just want more #ChrisContent about music and the computational humanities, check out the most recent episode of the Café Technika podcast! podcasts.apple.com/us/podcast/e...

This morning I signed a contract with Routledge for my next book.... The AI Music Problem: Why Machine Learning Conflicts With Musical Creativity It's due to them early fall, so it'll probably be in print early summer 2025!

The amazing Patrick Hicks interviewed me on his substack podcast about music and AI! Check it out: patrickhicks.substack.com/p/why-does-a...

I'm giving the keynote "Why is AI so Bad at Music (at least so far)?" at the New England Music Library Association May 31st at Smith. Reg. is free for 1st-time attendees, so if you're looking for some hot takes on a Friday afternoon, come heckle me! nemla.musiclibraryassoc.org/nemla-meetin...

I’ve got a pretty hot take (for music theory) out today, coauthored with a veritable flamboyance of amazing scholars! #musicScience #musicTheory mtosmt.org/issues/mto.2...

How much much smaller are musical datasets than those in other media? This lil infographic shows how long it would take an immortal human to consume various content sources. More curated and accessible musical datasets are quite small compared to what big-brand LLMs use! #musicScience

Workshopping titles for the new book:

Here's my first foray into music & AI! It's about interpreting Generative Pretrained Models (GPTs) by mapping internal connections under the hood, and by stopping its training periodically to see when/how it "learns" #musictheory concepts tinyurl.com/bdfvuya4 #musicscience

I was writing about the myriad reasons for why musical datasets are so small, and just wrote the sentence, "There's less cheesecake in the world than there is cheese" and I think I'll never construct a better metaphor

I've been doing some research into early critiques of AI and it's FASCINATING. In 1842, Ada Lovelace wrote the first computer code. In her notes, she used a #music example to talk about AI's potentials, and in the same breath talked about its limitations. She wrote: (1/x)

doing a little back-of-the-envelope math here while working on a history of #Music #AI: a computation that takes my current laptop 2 minutes to complete would have taken David Cope’s or H.J. Maxwell’s computer over 1 year of uninterrupted work. #musicscience

Here's my piece for New England Public Media! www.nepm.org/commentary/2...

any of you listening to New England Public Media, I'll be on tomorrow (Wednesday) morning a a few minutes before 6:30 and before 8:30!

Scriabin legos:

PSA: Writing requires SO MUCH EFFORT. I was reminded of this when I sat down to edit my own prose (of what I thought was a near-final draft), and ended up with pages that looked like this. A prof once told me 'writing that sounds effortless takes a lot of effort'. Solid #truth

One of my students put together a survey to see whether folks can tell which of two rhythms was written by AI... want to see if you can tell? We be grateful if you spent two minutes trying! docs.google.com/forms/d/14Ny...

AI: Can you make me a picture of a music theorist?

Strong recommend of Cher's new Christmas album. It's a beautiful glorious rainbow of a trainwreck, and it'll be on repeat in my ears for the next month

Wm. (Gosh, why did I let myself get goaded into that in 2013)

Preprints are up for our paper "Metric Accent Affects Perception of Key Center in Pop-Music Chord Loops"! We find that listeners use 2 strategies to assign key in loop-based pop... 1/2: osf.io/preprints/ps...

Just used #GPT4 to format my bibliography, and it inserted Taruskin's History into it. I do not cite Taruskin in this paper, but clearly this LLM thinks I should.

Minor horror from this morning: There's this popular corn maze in my town, & on my morning runs I sneak in through a back entrance I know & use it as a sort-of track. As I turned a corner this morning, standing was standing at the other end of the maze's corridor. I turned and got out quick.

doing my yearly listening of "Today's the Day" from Reply All (2015), perhaps one of the most poignant piece of podcasting that I'll listen to every August. And it's just made more poignant every year by what's happened since then. @pjvogt.bsky.social @agoldman.bsky.social

I ordered some William Grant Still music, and I’m loving that it comes packed with antiracism literature, pages of which are clearly copies of copies of copies from decades ago