[MUSIC - Breaking Rust: Walk My Walk]

Host: No, that's not the Brian Lehrer Show theme music. That's Walk My Walk, the number one hit on Billboard's country digital songs sales chart. Say that five times fast. The performer behind the song is Breaking Rust. Or at least that's the name of the avatar behind the song. This hit is one of several AI-generated songs to have reached broad audiences this year. Earlier this year, a band called Velvet Sundown garnered over one million streams before its creators admitted that the band's song was created with AI. Earlier this month, an AI R&B singer, Xania Monet, the product of a poet named Talisha Jones, landed a multi-million dollar record deal after topping the Billboard charts. Songs generated using artificial intelligence are dominating top music charts and streaming services like Spotify, sometimes without the knowledge of listeners.

Joining us now to break down how we got here is Kyle Chayka, staff writer at The New Yorker, covering technology and internet culture. Kyle, welcome back to WNYC.

Kyle Chayka: Thank you for having me again. It was such a treat to hear that song.

Host: Had you heard it before?

Kyle Chayka: Oh, yes, for sure.

Host: Listeners, I wonder if any of you have heard that song or any other AI-generated songs lately. Maybe it was pushed to you on your Spotify weekly playlist. Maybe you heard it on a social media platform and decided to look up the artist just to realize the artist is an avatar. If you've come across AI songs, how did you know, and how did it make you feel? Or maybe you're a musician or a songwriter, or a producer. What's your take on AI music? Is the barrier to entry too low, or does the disruption maybe equalize the playing field?

212-433-WNYC. That's 212-433-9692. You can also text that number.

Kyle, a few years ago, when AI music was just coming out, whoever was generating the songs was replicating well-known artists. I think you mentioned you're reporting Drake and Jay Z, and I remember it sparked a conversation about rights at the time and intellectual property. You write now we're in the midst of a full-blown AI music moment. What has changed?

Kyle Chayka: It's a good question. I think those early tracks went viral because they were so closely mimicking these pop stars we were familiar with. It blows your mind that you can hear a fake Jay Z voice, and it sounds just like Jay Z. Now, these AI tracks, they don't have a famous voice behind them. There's nothing to recognize; there's no name recognition or identity behind them. It's just this anonymous avatar situation. I think it's happened in part because AI music technology has actually gotten even better from a few years ago.

Host: In your latest for the New Yorker, you profile Nick Arter, a 35-year-old in Washington D.C. who started making AI-generated music late last year and has since netted, you say, hundreds of thousands of plays. Can you walk us a bit through his process? What software does he use, and to what extent does he really control what gets spit out on the other end?

Kyle Chayka: Yes, Nick has had a really fascinating career because he's been making music and rapping almost his entire life. He has this deep musical background, but he's never been a professional. Then, about a year ago, last November, he started experimenting with these two AI music apps called Suno and Udio. Those are the two main options that people are using right now.

What he does is he writes all the lyrics himself. He sketches them down in his phone or writes them out on weekends, and then he'll feed those lyrics into Suno or Udio and then describe the track he wants to make, like the genre or the year, or the vibe or emotion that he wants to get across. Then the generative AI will spit out an entire song. It'll come up with the melody, the instrumentation, the recording quality, the singing. The lyrics are still Nick's. It is still a mix of human and AI effort.

Host: Kyle, we got a text right away at this segment, said, "Hello, my place of work-- Oh, it just disappeared from my screen. Here it is. "Hello, my place of work loves AI-generated music, specifically Nick Hustles. I looked up who Nick Hustles is, and that's-- Go ahead.

Kyle Chayka: That's my guy. That's Nick Arter.

Host: Nick Arter. Very cool. You write for aspiring musicians; this new ecosystem can feel bracingly democratic. What do some of these creators believe about the music industry and maybe the impact that they're having on it?

Kyle Chayka: At least according to Nick, he felt really stymied by the gatekeepers of the recording industry. He wasn't able to get the attention of labels. Making music with AI and just uploading it to streaming services gave him this access to an audience that he never would have had otherwise. Now that he's proved his success, record labels are knocking on his door, and he's hesitant about going with them just because he's done so much on his own.

I think the combination of generative AI and the wider streaming ecosystem has just made it possible for almost anyone to make music and then put it out there. That's not to speak to its quality. I interviewed other AI musicians who have not gone viral and have not had that success. You can make music and get listeners without being a musician in a way.

Host: On the topic of quality, since you raise it, you write, popularity of AI Music notwithstanding, AI is not, by most metrics, a good songwriter. I do want to use this as an excuse to play a stretch of the AI R&B singer Xania Monet. I'm not going to say that this is good or bad songwriting, but let's take a listen to one of her songs, This is How Was I Supposed to Know which I will just note has 6.7 million streams on Spotify as of this morning. Let's take a listen.

[MUSIC - Xania Monet: How Was I Supposed to Know]

Host: All right, Kyle, so you didn't single out this song in particular in your reporting. I don't know if you want to give us a hot take, if you've heard it before, or if you maybe want to talk about what some of the experts said about AI songwriting and what AI gets wrong in songwriting.

Kyle Chayka: It's fascinating. When you listen to a track like that, it has this human, emotional, expressive quality because you feel like the voice is making so much effort, but it's maxed out. It's like no human could actually sing like that. It is blatantly AI-generated. The lyrics, the poet behind that singer could have written the lyrics. It could be AI-generated. I think the musicians and commentators I spoke to noticed that there was this hollowness or tinniness to AI music. It doesn't ultimately feel like there's a lot of soul behind it.

I think it's too perfect, it's too maximized for our listening pleasure. It's too hooky. It's a reverse-engineered candy for your ears. I just feel like there's not enough messiness, like human messiness behind it to be that interesting.

Host: Listener texts, "If these songs are generated by AI, then how can you be sure that the listens aren't computer-generated as well? Is there any sense of how organic this phenomenon is?" Kyle, do you know?

Kyle Chayka: For sure. Talking to Nick Arter or Nick Hustles, he says that all of his traffic is totally organic. He has never put in any effort to marketing or putting his stuff out there in different channels besides YouTube and Spotify. For him, it's totally organic. I have seen other reporters note that Breaking Rust, the Walk My Walk, AI musician that seems to have a lot of bot traffic. I think with Nick, you can see the listeners are there, the commenters are there. As one of your listeners just said, they play his music at their workplace.

I think there is an organic popularity to it, and there is an appeal, and there are some level of genuine listeners. My analysis of it, I think, is that we're so used to algorithmic recommendations and these passive feeds of content online that we just don't always notice that it's AI, and so probably a lot of people are just listening to it and not thinking about it too much.

Host: Or Kyle, maybe the text message about listening to AI music was AI-generated itself, as maybe is to be expected. We are getting a lot of texts from musicians and people who work in music. Here's one. "I'm an old school pro jazz musician drummer with also a background in pop and rock and roll. I'm horrified and disgusted by AI infecting music. For me, music is about soul and spirit of human being connecting with an audience. Being 62, I have had enough trouble accepting the inevitable arrival of MIDI. This is a whole new level which I find completely unreliable and terrifying."

The last question here is the infecting of AI in so many aspects of daily life so quickly is a huge error and not welcome trend. Do you welcome a trend to usher in by a group of isolated self-obsessed billionaires? I don't. Rhetorical question there. Let's go to a quick call. Susan, in Summit, you're on WNYC. Hi, Susan.

Susan: Hi. Oh, I'm so glad you took my call.

Host: Thanks for calling in.

Susan: Anyway, I just wanted to say that I'm so fed up with this AI. My daughter has been struggling for years and years to get into the music industry. It's not easy, no matter what. Now we have this AI stuff, which is not real, and I'm just fed up. What happens to real people? Why is this considered such a fantastic thing that's going on? It just gets me angry. I don't know what else more to say. The gentleman seems to think that it's spectacular. Maybe I'm wrong, but no, thank you.

Host: Thank you, Susan, for your call. Kyle, I'm not sure that that's what I got from your reporting, but a really interesting wrinkle that Susan adds there. If Nick Arter says, "This is democratizing music and lowering the level of entry," Susan is saying, "Well, wait, not for real." Now the field is even more crowded. What do you think?

Kyle Chayka: There's more competition. I think it's so easy for new people to get into the marketplace. Let's say that it does feel like there's no barrier to entry at this point, and that does make it harder for humans to succeed. I don't think that's spectacular by any stretch of the imagination, but I was interested to understand from these musicians that they do feel creative. They do feel like they are producing something that has value, and a lot of listeners are consuming it on some level as well.

I think the real culprit here that's really aggravating is that Spotify just doesn't identify which songs are AI augmented. We have no choice. We can't tell ourselves if AI has been used. We're left to analyze each one and decide if we think it's human enough or not. I think a lot of people, if they could just flip a switch to hear no AI music ever, they would, myself included.

Host: I think that, as you note, Spotify is improving its AI filters without defining what qualifies as an AI song. Do you want to briefly untangle that for us? Are there any definitions of AI songs or not yet?

Kyle Chayka: It's a bit unclear. As one of the listeners noted, MIDI technology was seen as a disruption of music as well. Now we have AI-influenced music. I think we're in a social debate or collective debate over what is human enough, what qualifies as human, because auto-tune is also a form of AI, and plenty of music uses that. I do think, I don't know, you know it when you hear it, when something is too automated and too technological. I think Spotify is struggling to find that line right now because a lot of musicians are using AI, and then it's up to the platform to decide what is too much.

Host: Kyle Chayka is a staff writer at The New Yorker covering technology and internet culture. His latest in The New Yorker is that new hit song on Spotify, it was made by AI. Kyle, thanks for joining us.

Kyle Chayka: Thank you. This was fun.

[music]

Copyright © 2025 New York Public Radio. All rights reserved. Visit our website terms of use at www.wnyc.org for further information.

New York Public Radio transcripts are created on a rush deadline, often by contractors. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of New York Public Radio’s programming is the audio record.