2025-08-30
Quick note: I’m not a fan of including substantive tangents in footnotes because I think it really does mess with the Reading Experience, or at least mine, jumping back and forth in the text. However, there are so many asides and caveats and additional notes I want to make about this topic that I am making the executive decision to Subject You To It a few times. Feel free to read the whole thing first, then dive into the notes—or not touch the notes at all. Whatever brings you the most joy!
Dear friend—
In Mockingjay, the third book in the Hunger Games series, the main characters play a game. Peeta has been literally brainwashed by the Capital, and his hold on reality and his memories has been ruptured. So when he remembers something of Katniss, he asks her, “Real or not real?”
I, luckily, have not had my memories tampered with by a blood-thirsty authoritarian regime, but I find myself thinking of this refrain more and more nowadays.
Until recently, I loved selecting 3-hour-long YouTube videos of curated playlists to accompany me while I write and work. A lot of these seem created for the express purpose of playing in the background while you do something else—they’re instrumental, they’re Bossa nova jazz, they’re “oldies playing from another room,” they’re “10 HOUR Ghibli Classics on Piano NO ADS” etc. etc.
About six months ago, I found out one of my (former) favorite of these videos is straight-up AI-generated music.
YouTube’s catalog in this genre is now teeming with AI. “Jazz to study to” was one of my most oft-searched phrases, and the results are now livestreams with clearly AI-generated art and maybe-AI-generated music. And part of the problem, for me, is the not knowing. Often there’s no big label declaring, “MADE BY AI!” I’m left to comb the comments for other folks who have caught it first (and they might not even be correct on that count).
My default has become suspicion. If the video doesn’t include a track list with an artist or two I recognize (because even random named artists on Spotify are turning out to be AI-generated), I don’t listen to the video.

My qualms with AI-generated content are many—from the environmental impacts, to the present and future labor issues, to the provenance of the data and artworks that the models are trained on.1
But the thing I’m thinking about more and more is the way AI-generated content is fucking with or will come to fuck with our understanding of reality. How the world around us will become a funhouse of representations of the Physical Plane that may or may not have human creators or validators and will essentially be robot hallucinations.
With AI-generated music, the stakes are relatively low. In the age of Ticketmaster and Spotify turning the music industry into yet another “Winner-Takes-All” bonanza, where only the richest of artists can make a living from making music, I sense a growing shared value amongst my peers to support indie and small artists. We WANT to subscribe to the Patreon, the Substack, the SoundCloud.
Perhaps this is simply my naïveté, but when I realized the amount of AI-generated music I was listening to, I felt betrayed. Rather than supporting an indie artist, I was contributing ad revenue to someone’s get rich quick scheme.2
But the issue goes beyond the simple question of “Bot-created or human-created?” The generative AI of today—from ChatGPT to deepfake videos (videos that look like they’re filming events that happened in the physical world but were edited or completely AI-generated to show something that never happened) seems dangerously unconcerned with the truth, the reality of the world, the bare facts of y = mx + b. And the more I learn and think about this, the more I feel we are headed down a dark path.
.
I keep hearing this argument regarding the fact that AI search is mostly accurate but not all the time. “It’s fine to use it if the question you’re asking is low-stakes, but skip it if accuracy or rigor really matters, like with research.”
UH don’t you want the right answer all the time? Or as much as is feasible?
Sure, I won’t get a single perfect, infallible answer from anyone or anything on the internet for questions such as, “Is God real?” or “Is communism evil?”—the kind of stuff that is pretty subjective and swayed by what sources you look to and requires poring over great amounts of media to form your own thoughts about.
But like—“How do you replace the filter on this model refrigerator?” “How many cups are in a gallon?” “What year was Cher born?”-type shit—Don’t we want to just have facts for the sake of having a shared understanding of the universe? Of knowing things to be true?
Of course, the margin for error on the internet was large before AI chatbots and search functions. Nevertheless, I don’t want to look up some random fact and have an 80 or even 90% confidence that the answer This Robot gives me is correct. That sounds like the road to hell! And perhaps more importantly, I want to know who is giving me this information and have some trust in the relaying of information > informer > me, the informee.
With tools like Google’s AI search and ChatGPT, this chain of knowledge is getting thrown by the wayside, delivering us text that sounds authoritative but could be a complete hallucination—and perhaps even the sources the bot cites are hallucinations in and of themselves!
.
Recently, I listened to a podcast interview with the CEO of a company whose platform allows people to create deepfake videos. He imagines aloud what he calls the worst-case scenario for a society where deepfakes are just really, really accessible:
I think the most likely scenario is that we kind of go back to how things were 100 years ago. Essentially, video didn't exist. Audio recordings didn't really exist. Photos barely existed. And so let's just call it, they didn't exist.
Everything was practically just a matter of trust. Like what you believed and what you saw just depended on, Oh, well, who wrote it?—Well, it's this publication, the XYZ publication, and they're very reputable, you know? …
Like nothing can be actually recorded. So the only thing that can be recorded is what people have seen. And that's the only thing that can be trusted, but that can be trusted only as much as you trust the person.
The CEO doesn’t seem too bothered by this. “We’ve done it before, we can do it again”-type attitude.
Which, to a point, like—fair. Perhaps humanity would continue on in some relatively-okay-somewhat-healthy form if we just went back to the days of newspapers and reporters and institutions. Maybe it would be bad in ways that we’ve seen before, not an “existential threat to the whole of the species” bad.
But also, I think the big gaping hole in that perspective is that we don’t live the way we did 100 years ago. We’re not at the starting line, today, with a particularly strong shared sense of reality.
The digital environment has already put us into little boxes where our sense of self and the world is greatly circumscribed by the people we follow, the outlets we place our trust in, and the algorithms that our tech overlords deploy to show us things that we will really love or really hate.
Meanwhile (as I often need to remind myself), there is a vast populace of people who are not chronically online and don’t even know or care about this stuff, because it’s beeboopbeeboop and their lives are solidly in the Physical Plane and they only get their news from their twice-weekly Facebook check-in or from their coworker chatting about it at work.
For many, the “Real or Not Real” game is far, far down the list of Matters Of Concern. It ranks below things like how to grow tomatoes from a seed or how to stop a toilet from overflowing or how to take care of an incredibly ill person or how to survive our capitalist hellscape from hour to hour.
And that last bit—perhaps that’s what makes the rise of deep fakes and AI-generated content even more insidious. AI companies seem to be luring us into a confused sense of reality and truth at the exact moment when everyone is either overstimulated by a constant wave of digital media or barely keeping their heads above the churning waters of our current dystopia.
All this to say, I have little interest in winning or losing the “Real or Not Real” game. I don’t want to invest my mental energy into learning the tiniest tells for AI-generated content that will shift in the next month as the models learn and change. Nor do I don’t want to mindlessly inundate my brain with Spurious Robot Hallucinations.
And I don’t want to get even a little bit dependent on yet one more piece of technology that may irrevocably rewire my brain to be less capable, less critical, and less in touch with the world of flesh and bone and grass and sky. My smartphone does that enough on its own.3
I do feel very mixed about this bit because I think uncritical praise of the analog and “how things used to be” can veer toward snobbishness and conservatism.
But I’m gonna be real—the more the digital world feels batshit, the more I feel the need for groundedness and being more in touch with the World Before Me and everything in it that allows for art and connection. So here are a few examples of ways I’m trying to arrange my aural entertainment/environment to that end.
The benefit of these is that I can Trust Their Provenance—plus, there’s a kind of freedom in not being beholden to Big Tech’s algorithms and instead enjoy my own tastes and another person’s (which, to be fair, are still algorithmically influenced likely to a Big Degree, but at least Mark Zuckerberg’s Own Bots aren’t immediately crawling around my grey matter).
Recently, I’m been more often:
Listening to live performances of my favorite artists. Live music really does hit different. I love the energy of the crowd, I love the new directions the artists take their songs, I love the roundness and energy in the sound. This concert of Bleachers at Radio City Music Hall, this concert of Declan McKenna at TRNSMT, and this live album from Dijon are often on repeat.
Falling asleep to the radio and waking up to it as my alarm. There was a jazz oldies station in my Idaho town that I woke up to every morning, but there’s no equivalent here :( I can stream it on my phone, but (at the risk of veering further into Old Man Yells at Cloud territory) it really isn’t the same.
The closest thing I’ve found in my current place of residence is the local public radio’s two-hour jazz program in the evenings, right around my bedtime. So I’ve been trying (trying being the operative word here) to end my nights with a bit of reading and radio. This also has the benefit of allowing me to charge my phone very, very far away from me.
Starting my mornings on my record player. Right before I moved to Idaho, I found a little record player at a garage sale new-in-box for $20. It’s very “Electronics-Section-At-Kohl’s” quality, but it gets the job done.
I didn’t take it to Idaho, so I’m very glad to be enjoying it now, and the only three records in my collection currently are greatest hits compilations for Duke Ellington, Dinah Washington, and the Mills Brothers. There’s a used record store in my local mall, so I’m excited to visit again soon!
And I’m still doing work to instrumental music hosted on YouTube, as it’s my go-to way to Get My Head In the Game at my laptop job—only I’m taking extra care to ensure it’s indeed human-created and sticking with my faves, in order to reduce my “Real or Not Real?” experiences.
For something more upbeat I like this collection of Saxophone City Pop covers by sanpond and this FUSION JAZZ playlist; for times when I need an aural environment with more thinking space, I’ve been digging “peaceful frog music to zone out to” by Tanger and camiidae’s “cambrian animals” album. And, of course, typing ASMR videos (real hands on real keyboards) never fail me (though who knows, maybe the deep fakes will come for them in time :/)
Would love to hear what things you’ve picked up to help you feel more human in our robot age! And always down for recs of more playlists to add to my playlist of playlists :)
Thanks for reading, chat soon <3
—mia xx
I do think the term “AI” has been flattened to the point of having nearly no meaning. Putting technology that optimizes the grid, analyzes data for cures to disease, creates kill lists for armies, turns out AI slop like Ballerina Cappuccina, and tells you what year Cher was born all under the same label feels extremely reductive. So in general, when I say “AI” in this piece, I’m specifically referring to generative AI, the stuff of chat bots and image generators.
Also, my distrust and distaste for generative AI is based on AI as it is being deployed today—built on all sorts of exploitation, mostly in service of driving growing profits and reach and power for Big Tech monopolies than in Service of a Public or Common Good—than AI as a technology in and of itself or in its platonic ideal. Can AI be used for good or made without exploitation? IDK, maybe, some people are working on it, but that sure as hell ain’t the Dominant Mode at the moment.
Perhaps there’s a philosophical discussion to be had about whether prompting a chat bot to make the sounds you imagine in your head but can’t play on an instrument is still “making music.” I’m putting a pin in that—the answer for me is, I don’t know! I do think there’s magic that’s lost in the jump from “making songs off Garage Band samples and beats you pulled off YouTube” and “Robot spitting out music in response to some paragraphs you typed.” But maybe in 5 years there will be a serious argument for prompting as its own craft?? All this to say, I’m much more set on the economic and societal dangers of AI as it is now than the philosophical debates about creativity and craft surrounding it.
Of course, the question of AI’s roles in our lives often has far more to do with government and corporate decisions—like contracts that put generative AI tools in public school classrooms or bargain-bin deals offering ChatGPT to college students—that make opting out increasingly difficult. Preventing that sort of stuff takes political and collective, not personal action. But discussing that is a whole ‘nother essay in itself. Sorry y’all, I’m stopping here—but suffice it to say, personal choices + collective political action go hand-in-hand!
Don't miss what's next. Subscribe to accidental light:
Add a comment: