(This was originally published as Implausipod Episode 49 on July 7th, 2025.)
https://www.implausipod.com/1935232/episodes/17441034-e0049-spirit-of-the-ai-dio
A look into the rise of ghost artists on Spotify, both AI generated and not, and what the history of Performer’s Rights Organizations mean for art and creativity in the 21st century, and how that may make us question the very nature of creativity itself.
Let me ask you a question. What do you do if you’re a musician working the mean streets of New York City trying to get paid for your work? You see, you’ve made some compositions, but thanks to some hot new tech, anybody can copy it and hear the songs, the music that you wrote, and you don’t get paid a single penny and New York City isn’t cheap.
It’s rough for a musician to make it, but this new tech, and you’ll admit it is a marvelous invention. Makes it hard for you to make a living. But the tech does have its limitations. It’s easier to copy and share your tunes for sure, but they still need to be copied by someone transferred to media. That limitation, that drawback gives you a crack or maybe just maybe you can get paid for your music.
This is the situation Victor Herbert found himself in a little over 110 years ago, and we’re going to look at exactly what the ramifications of his solution was. In this episode of the ImplausiPod.
Welcome to the ImplausiPod, a podcast about the intersection of art, technology, and popular culture. I’m your host, Dr. Implausible, and this episode has some links, not just to stuff that we’ve been discussing here, but to some recent events in the news, and it’s gonna take some twists and turns. You see the solution that Victor Herbert and some of the other composers in and around New York City came up with in the early nineteen hundreds to help solve their problems has a lot to say about the current state of media in 2025.
You see in the development of a new technology, a lot rides on the physical limitations of the media. Often that could come down to logistical, practical concerns, the ease of duplicating something or transporting it. What Victor Herbert was dealing with was the rise of rolls of music for player pianos, the hot new tech at the time, tech that could be copied and shared and meant that he was losing opportunities to get paid for playing it.
So when Victor and a few of his fellow composers on Tin Pan Alley got together to perform the first. PRO or performance rights organization, one that would negotiate collectively on behalf of its member artists is ASCAP the American Society of Composers, authors and publishers. What they ended up doing, whether intentionally or not, for both ASCAP and the other PROs that follow, was providing a means for listeners to address some of the ethical concerns that they may have had when it comes to the content that they were consuming.
Hmm. It sounds a little weird when it’s phrased that way, talking about listening to music in 1914, in 21st century terms, but that’s basically what was going on, and that’s why the story of how the PROs came about is relevant to us today too. In one of our recent podcasts back in episode 42, where we talked about the incipient diaspora of the potential end of TikTok, we discussed how making informed choices and ethical consumption matters when it comes to media.
At the beginning of our episodes, we sometimes mention that we’re not on Spotify, and that this is an intentional act. I’m not a fan and I don’t like their business model, so I’m not using them. In late 2024 and 2025, some news came out about how Spotify was using AI generated content, algorithmically developed for easy listening and the reasons why we’re not on Spotify became crystal clear.
Now finding there’s a market for endlessly looping smooth jazz isn’t that surprising. It’s a concept that became so ubiquitous that a word was coined for it. Muzak. Invented in the 1920s by American George Owen Squier, Muzak was a non-radio form of music delivery that used the electrical wires to deliver songs directly to paid subscribers over the air. Systems like radio were inconsistent and spotty at the time, so there were takers for this new system. Think of it as an early version of broadband over electrical that you could set up in your own home today. Once radio started to catch on with the home market, Muzak shifted to business customers and as the company changed hands and ownership, it was used to regulate the mood in the environment where it was delivered.
A fast pace equals faster workers, or so the Taylorist line of reasoning went. Muzak was peak in the 1950s and sixties, but gradually became to be associated with bland corporate music, as competitors licensing more popular music came on board providing similar services. It would take a few years still for the popular music to also become bland and corporate.
But I digress. By the time the competitors started appearing, Muzak had become a genericized trademark like Jello, and it doesn’t really make a difference what version of elevator music you end up hearing, just that you’re hearing it. Which is where Spotify comes back into the story.
I said the endlessly looping background music isn’t that big of a surprise. How they are generating it as the use of AI for the delivery of muzak represents a sizable shift. And so in this episode of the ImplausiPod, we’re looking at the spirit in the machine, or in this case the spirit of the AI-dio. And here’s where we’d queue up that Rush song, if I had a budget for music licensing, or even for the muzak version. I’ll trust that you can hum along.
Now one sure thing about studying work in the AI space is that it moves incredibly quickly. It is acceleration made manifest, moving at a ridiculously quick speed. This velocity can be sensed, almost felt giving your eyes to the feeling of an ease many have when dealing with it. That and the killer robots, which we discussed earlier.
Of course, I say this as I started writing this episode back in December of 2024, based on a few articles that I had read, and a then forthcoming book, which came out back in January. Since then, the conditions being described progressed substantially in new stories were continually being added to the topic.
It turns out I have a bit of a halting problem when it comes to researching these episodes. Some of the things that we were planning on talking about have come to pass and we’ll. Still get to them, even though this episode will feel slightly less prescient now than it would’ve back in December. But cest La vie, it’s also a reminder that these things will always be like trying to hit a moving squirming target.
One of the ways to deal with the limit of this snowball sample that we’re working with is through a concept known as saturation. When new queries are not drawing in noticeably new or different information, you can stop the work and get to it. So now that we’ve drifted enough from the original topic, let’s do exactly that.
In December of 2024, the blogger Ted Gioia published a piece about The Ugly Truth of Spotify on his Honest Broker blog, and that he walked through the observations he was making about jazz playlists filled with artists he hadn’t heard before. They were also musically identical tracks published under different names. It would keep showing up.
It’s not a big deal if it’s in the background of an office or a retail outlet like this often was when no one is looking too hard at the playlist. This is something that Spotify called PFC or Perfect Fit Content, which had a royalty rating that was favorable to Spotify. This work by Gioia coincided and resonated with the work that was being done by Liz Pelly, and he mentions her in his blog post, in her book on Spotify titled Mood Machine.
She was talking about the rise of ghost artists, something she had been tracking since 2017. This is a rumor where Spotify was quote “filling its most popular playlists with stock music attributed to pseudonymous musicians” end quote, much like the Muzak corporation of 80 years earlier. The thought was that Spotify might be making the tracks in-house, all in an effort to lower royalties in a market where streams were already fractions of a cent. And perhaps this is the moment where a little background on Spotify is an order in 2025. It is a ubiquitous brand name for streaming music, but it had to start somewhere.
Spotify is a Swedish online services company specializing in the delivery of streaming audio. This includes music as well as podcasts and audio books. Founded in 2006, it experienced rapid growth starting in 2011, and by 2015 had become the defacto streaming app on most platforms. With this growth, Spotify is now in position of being one of the key drivers of the music industry, setting rates in the business model that others must compete with.
And make no mistake, there are competitors. Tidal, the Swedish streaming service acquired by Super Bowl impresario, Jay-Z in 2015, and subsequently sold to ex-Twitter honcho Jack Dorsey currently has market share, and the now venerable iTunes from Apple still accounts from about 12.6% of the market share as well with Amazon and Google’s own YouTube music falling at 11.1 and 9.7% respectively.
So Spotify isn’t alone, but the scope of their business worldwide is staggering. They announced that the payouts they made to the music industry was in the neighborhood of $10 billion in 2024 alone, and that year was also the first year that it was profitable, providing those payouts from revenue of $15.7 billion.
But not all is rosy in Spotify land. Aside from the outsized influence they wield on the music industry, which would be bad enough in and of itself, Spotify has been the subject of controversy for almost its entire existence. Most prominently is the pay rate that they give out for artists, which can be about 0.0029 cents per stream. For your mega stars with millions or billions of streams, your Taylor Swifts and the like, this can still amount to a decent return, but it falls off rapidly. One would need about 1.7 billion streams if my trusty calculator is working correctly to earn the median income in the United States if one was being paid at that lowest rate. Though the rate does go up to an average of what Spotify states is about 0.70 cents per stream according to their press releases.
So over 10,000 artists make a hundred thousand dollars or more using their streaming services, but. Of course many artists earn much less than that. Spotify operates on the classic long tail model where a minority of artists make an outsize amount of the revenue, and most of the rest gets a tiny fraction of the sales. This business model can be seen in many cultural industries like the movies, book, sales, traditional music, and even things like OnlyFans. One or two big hits ends up funding the label or a platform, and the others break even if they’re lucky or more likely are a loss. This is ultimately a speculative enterprise, at least how it is constructive in the capitalist framework.
And this speculation preys on the artists as well, where dreams of quote, making it big” provide a constant stream of new entrants to the industry. This never-ending flood of new artists and content has been why the CEO of Spotify, Daniel Eck has said on record on Twitter in 2024 that quote “the cost of creating content was close to zero”.
Or sometimes less than zero, as much of the expenses of music production are born by the artists, and even after all that effort, they may not recoup anything if they list on Spotify. In November 2023, Spotify announced that they would no longer pay artists for less than a thousand streams, effectively cutting off many small artists from earning any income whatsoever from the platform.
And the list of Spotify’s misdeeds grows from there. While cutting off small artists from revenue, they turn around and take those funds to finance high profile artists like Joe Rogan and others. And recently Spotify CEO Daniel Eck made the headlines for a billion-dollar investment in drone warfare company Helsing, of vampire hunting fame. A German defense contractor, which uses AI for the control systems in its aerial and underwater swarm drone technologies.
They also create a virtual environment, which provides the drones with spatial awareness, and we’ll look into that in a future episode. Their technologies are currently being actively used in the Russo-Ukrainian War. Ek’s investment has caused an uproar among some Spotify users with cancellations being directly attributed to that connection and investment.
And of course, along with all that, there’s the aforementioned PFC. Depending on the extent of it, Spotify may be one of the few companies turning a profit on AI-fueled content. There’s no reliable measure on the extent of the issue, though it has been going on for years, and finally the amount of AI generated titles reached the point where it was noticeable to the keen observer, if not perhaps to the casual listening audience.
All of these reasons and a few more besides are why you can’t find the Implausipod on Spotify. Like we mentioned earlier, it’s an intentional act. When podcast creators say that they’re available everywhere or on all platforms, and they’re saying that the issues with the platform don’t matter to them. There’s a degree of what I like to call platform illiteracy going on, but we’ll save that topic for a later date. The end result of these developments with ai, music generation and algorithmic delivery is that we are now living in a world with endlessly available, unique instrumental music. So much of it is being created that you could listen for a lifetime and never hear the same song twice.
Now, this is also technically true under the current model with 120,000 new tracks hitting Spotify every day according to a 2023 article by Maurice Schon. But again, our focus here is on the AI generated music.
Hold that note in your head, that little snippet of the interstitial music we use for the show. We’ll get back to that in a hot minute. We need to address the question at hand. What’s the problem with AI generated music anyways, about six months ago, there’s a trend of AI style covers playing Metallica and the style of a fifties doo-wop band or whatever. And while that was an amusing exercise, the novelty soon wore off. There’s only so much of that kind of act that you can take as Richard Cheese and Me First and Gimme Gimmes can well attest. Clearly that kind of style cover or genre switch can be done without AI, but all the transformers are doing is accelerating the process, filling some niches that otherwise might never get explored.
If AI generated music is filling a need there, and otherwise it’s mostly supplanting the niche previously occupied by Muzak for inoffensive background noise, what’s the issue? Perhaps the issue is quote-unquote “authenticity”. I say that because literally, as I was in the middle of recording this, the news story came out about a hot new band on Spotify called Velvet Sundown.
They play a radio friendly mix of seventies rock and indie pop, and they had amassed over a million monthly listeners when people began looking to see if there’s more info, because it’s not like music fans are the ones to become obsessive about their favorite band. And what those music fans noticed was something that had a lot in common with the music noticed by Liz Pelly and Ted Gioia earlier.
Odd connections and inconsistencies and a lack of the data or digital footprint we’d expect to see of a band if they had been around for a while. It now looks like the band is a complete fabrication with AI generated art and music. A man operating under the pseudonym, Andrew Prelon, claimed responsibility saying that the music was generated with Suno AI and that the whole project was a quote unquote art hoax.
But even that might be in dispute as there’s more than one claimant that says they’re acting on behalf of the band. It may have been that there was another AI artist out there, and Prelon just decided to step in and act as the band’s publicist, and that little bit of the hoax was completely tangential to whatever was actually going on with Velvet Sundown.
What Prelon and the Velvet Sundown affair highlight is the question of whether a producer of an AI art is actually the artist. They’re the driving force, commissioning the various elements of the work. If so, do they occupy a similar role to managers of boy bands like Lou Pearlman and the Backstreet Boys and NSYNC, or Malcolm McClaren and the Sex Pistols?
At some level, these bands are still quote unquote authentic, even though they’re clearly manufactured in the same way that a chipboard table from IKEA is still a table in form and function, even if it’s not handcrafted from oak. This authenticity of art is one that has been under scrutiny since the dawn of the 20th century.
Walter Benjamin discussed how art loses its aura in an age of mechanical reproduction, where the aura is the very thing that cannot be reproduced. But maybe this whole Velvet Sundown thing highlights the way. If the music is replaceable, then maybe the art lies elsewhere.
When attempting to answer all these questions, much of it comes down to the position one takes on AI ethics. This is often driven by our feelings. The way AI ethics is framed in the media often leads one to believe that the only ethical stance is to oppose its use on all levels, and we see that cropping up more and more.
But this often feels like taking sides in a battle between billionaires, just as the image we have in our mind of the small independent farmers, often exploited by agribusiness concerns, The mental image of the struggling artist is often leveraged by billionaires and IP rights holders. If we recall that Robert Downey Jr. has a net worth of around $300 million. We can perhaps understand his stance when it comes to AI generated arc, but for others, the position is less clear. And as we’re talking about songs here, perhaps we could focus on the music industry. The history of the music industry is rife with abuse and exploitation where original artists have been tricked, coerced, or threatened into signing away the rights to the music that has gone on to make others millions.
By way of example about what copyright can mean for artists at the time of recording, the Verdict is being laid out in the trial of Sean Diddy Combs an artist who still pays Gordon Sumner AKA Sting, $2,000 a day every day, 365 days a year for the unauthorized use of a sample on “I’ll Be Missing You” in 1997.
At the time of his arrest, Combs had a net worth of $400 million. Sumner has a net worth of over $500 million and Combs’ former collaborator, Jimmy Page has an estimated net worth of $180 million. These artists have not done poorly, and granted these artists are household names with enduring legacies, but much like the farming analogy above, when looking at it from a distance, appears we are caught up in a proxy war between billionaires.
We may not want to be simp for either side in this fight. What confounds that ethical calculation when it comes to modern music is that much of the industry operates as a form of rentier capitalism. This is where property is held without new investment and used to extract rents. The intellectual property, the stuff under control of the rentier in this case is used for value extraction and they’re not really adding anything new to the system.
The near endless ownership of IP can be seen as the enclosure of the digital media commons, where the AI companies turn everything into soylent culture fighting against the enclosure of the analog media commons by the old guard media companies operating under the established paradigm. So what’s the solution to this entrenched warfare between media, titans, old and new?
We’re not trying to rehabilitate Spotify. Rather, we’re here to adapt the idea of an artist’s rights organization for use in an age of generative AI. If we accept that there are valid uses for AI, and there are, we talked about this in episode 38, then there needs to be a path forward to dealing with this.
And as we hinted at in the beginning of the show, our friend Victor Herbert and ASCAP show us one of the ways that this might be accomplished, and there’s been some very recent moves forward on this front. The Creative Commons Organization has recently announced CC Signals, a licensing framework that will quote “allow data set holders to signal their preferences for how their content can be reused by machines based on a set of limited but meaningful options”. In addition, recent court cases have found that some of the data gathering done by the AI companies falls under the provisions of fair use. Together, these don’t cover every instance – it’s still early days – but it does show that there is a path forward out of this to something that’s equitable to the parties involved.
Of course, here’s the big twist, which probably wasn’t much of a shock if you parsed the punny episode title. There’s more than just the ethical question behind AI generated music. One that the AI-PROs may help ameliorate, but cuts us all closer to the core. We are seeing a great deal of Echange, of technological replacement, come to the music industry.
For musicians finding themselves replaced or that an algorithmically generated smooth jazz music act is good enough in a lot of instances, does this call into question the very nature of art and creativity itself? This appeal to creativity, the ad creo or ad fascia, depending on how my Latin is working, is something that has been called for increasingly during the debates around AI and the cultural industries dotting YouTube thumbnails and memes on blue sky and everywhere in between.
The ad creo is the claim that using an AI is anathema to the creative act, as if using a tool to generate an image somehow negates the spark and inspiration that led to the creation of the piece. This leads us to the Ditch Digger Fallacy. The counter to the ad creo of course is that what do you think creativity actually is?
Let me illustrate that question by an example. It has long been observed in nature that crows are particularly clever, that given sufficient motivation, usually a treat, they can use sticks or bits of wire to fish out a treat from within a piper or other closed environment where one wouldn’t expect the crow to be able to navigate at all.
This anthropocentric conceit of them having a limited bird brain refuses to let us believe what we were witnessing before our eyes. But even more complex behavior has been observed in crows. They appear to hold grudges. Yes, the birds got beef. And these grudges can both persist for years and be shared amongst the group.
Observations of crows engaged in group attacks gaining up on smaller animals or humans who cross them has gotten so bad that trackers have been set up in cities like Vancouver and Seattle to show the incidences and locations where the attacks have been fiercest. And the research is growing. The field of ethology is the study of the behavior and communication of non-human animals and has been producing fascinating findings that challenge our anthropocentric view of the world.
Much like the one we just mentioned, we are constantly finding creativity, communication, and intellect within the natural world. The more we observe it, and just like in other natural sciences, as the tools of observation improve, the more we can witness within nature. What we are seeing – what the ethologists are guiding us to – is that the more we can observe nature without disturbing it in some Heisenberg manner, the more we can observe the intelligence of the other species of life with which we share the planet.
And it leads us to ask, are we going to continually redefine intelligence as the ethologists uncover more and more ways that animals are smarter than we think they are? Is intelligence something anthropocentric, something we can only think of in human terms? If intelligence abounds around us in nature, in ways that were previously reserved for us in terms of problem solving, communication, emotion, grief, and so on, perhaps we’re not as special as we like to think, and this potential fills us with existential dread.
When it comes to creativity, perhaps our role is much more limited. Perhaps our role is that of the watchmaker, not the machinist building the gears. Recall the concept of the Allographic art that we introduced back in episode 38. This is the creation of art by other hands. The artist as architect or programmer, as choreographer or composer, the kinds of artists who Victor Herbert brought together when founding the first performers rights organization.
Here art is a question of control, and the skills in shaping art differ depending on the media. Within computing, one of the enduring tropes is that the users are like unto wizards and treating with demons in order to coax magic from the thinking sand. Here too, they must deal with the spirit of the AI-dio, the ghost in the machine.
Once again, thank you for joining us on the ImplausiPod. I’m your host, Dr. Implausible. You can reach me at drimplausible@implausipod.com, and you can also find the show archives and transcripts of all our previous shows at implausipod.com as well. I’m responsible for all elements of the show, including research, writing, mixing, mastering, and music, and the show is licensed under a Creative Commons 4.0 ShareAlike license.
You may have also noted that there was no advertising during the program and there’s no cost associated with the show, but it does grow from word of mouth of the community. So if you enjoy the show, please share it with a friend or two and pass it along. There’s also a buy me a coffee link on each show at implausipod.com, which will go to any hosting costs associated with the show.
