How AI is changing the music business

The world’s largest music streaming service now lets users to monetize music in which they don’t play or sing a single note. How is AI shaping how we make and profit from music?

Guests

Rick Beato, musician, engineer, and YouTuber.

Amanda Hoover, senior correspondent with Business Insider.

Also Featured

Jonathan Wyner, professor of music production and engineering at Berklee College of Music.

Transcript

Part I

CHAKRABARTI: Rick Beato is the consummate musician and consummate music teacher. He’s also an engineer and music educator whose YouTube channel Everything Music, which is @RickBeato. It’s racked up more than 5 million subscribers. He’s known for dissecting songs like Smoke on the Water or Mr. Brightside, and for interviewing musical legends like Joe Perry of Aerosmith and Billy Corrigan of Smashing Pumpkins, and he joins us now from Atlanta. Rick, welcome to On Point.

RICK BEATO: Thank you, Meghna. Appreciate it.

CHAKRABARTI: I’m a super fan of your channel. I just want to say I happened upon it when I watched your video when you dissected Kiss From A Rose by Seal.

And I’ve watched that video like 50 times. It’s so good. Was it fun making that? I just have to ask.

BEATO: It was very fun making it, especially since I got the tracks from Seal and then I didn’t even say Seal was going to be in the video, and then all of a sudden, there he is. I said why don’t I ask Seal about the lyrics?

And then he appears.

CHAKRABARTI: It’s just so good. Everyone, I really recommend that you watch it. I learned so much about that song that I didn’t know, even though I’ve listened to it, I don’t know, 5 million times, okay.

BEATO: It’s a really incredible song and very sophisticated.

CHAKRABARTI: Actually, we should keep that in mind.

It’s an interesting counterpoint for why we’re actually having you on the on the show today. You are here to introduce us to a new artist. The artist’s name is Lyra Vega. And let’s start by listening to a little bit of one of Lyra Vega’s songs.

(LYRICS) Streets are breathing under water blue. Every corner holds a trace of you. Footsteps echo where we used to run. Now I’m chasing shadows one by one.

CHAKRABARTI: Okay, so this is Neon Ghosts by Lyra Vega. What do you think? It’s pretty catchy.

BEATO: It is pretty catchy. (LAUGHS) It’s amazing what these platforms, this was done on Suno, which is one of the two main AI music platforms are Suno and Udo, and I created this, I guess you could say I created it with a prompt.

I actually prompted the idea of create a female, vocalist-led electronic pop track, and that was it. And it came up with the name Lyra Vega and the song title, Neon Ghosts.

CHAKRABARTI: Okay. Hang on for a second. Let’s listen to a little bit more.

(SONG PLAYS)

CHAKRABARTI: Okay this is AI. But if I didn’t know it and I heard this on the radio, I would honestly, I’d be like, okay, this is a up and coming Taylor Swift wannabe. But … how long did it take you to make this fake artist and song?

BEATO: Five minutes.

CHAKRABARTI: Five minutes.

BEATO: Yeah. It was essentially three prompts.

The first prompt was to create who the artist is, and I wrote, create an idea for a female vocalist led electronic pop song, what the song is about, and what the name of it is, and its artist. And then it created the Lyra Vega and the song title. And it had the concept. So the concept of the song is about wandering through a city at night after a breakup, and these ghosts are flashes of a memory, faces in reflections, laughter drifting from bars, and it goes through this whole concept of the story.

Then I went over to a program called Claude, which is Anthropic’s AI program. Now, I could have done this all in Suno. The actual AI music creation program. It will create lyrics, but I like to use different platforms to do this, different AI platforms. And then the prompt for that was create song lyrics in the style of an artist named Lyra Vega.

With songs sections denoted in square brackets about the following, and then I paste it in the prompt from ChatGPT.

And then it created the lyrics.

CHAKRABARTI: So wait, you said in the style of Lyra Vega, even though Lyra Vega doesn’t exist.

BEATO: Well the style and then I paste it in what ChatGPT said.

CHAKRABARTI: Okay. What the style —

BEATO: What the style is. Yes.

CHAKRABARTI: Okay. And so what exactly was the style that we just heard?

BEATO: I would call it electronic pop music, electronic dance/pop music, it reminds me of the song Clarity by Zedd and Foxes, or could be Katy Perry Firework, something like that, that has a dance beat with a four on the floor bass drum in the chorus. There’s plenty of pop songs that are like that. More, I would say probably back in the mid-2015 to 2020, you found these songs in pop music a lot more than right now. But this could easily be on the radio, on Spotify, on Apple Music, and people would have no idea it was AI.

CHAKRABARTI: A hundred percent. I wouldn’t have known unless basically you just said this to us. The platforms that you mentioned also created a face, a picture.

BEATO: That’s right. I actually asked afterwards, I asked ChatGPT to create me the face of Lyra Vega.

CHAKRABARTI: She’s blonde, I should say it, because she doesn’t actually exist.

It’s blonde. It has cat shaped eyes. Full red lips. Pretty much indistinguishable from a lot of, I don’t know, magazine photos of pop stars.

BEATO: That’s right.

CHAKRABARTI: Wow. And it says bold blonde and electrifying. Her style matches the high energy beats of her electronic pop anthems. Okay. So listen, you went to ChatGPT, you got a prompt and you put that in Claude for the lyrics. How would you rate the lyric quality of Neon Ghosts, perfumes off, like long pieces of clothing, is what I caught there.

BEATO: Yeah, they’re good.

Honestly, is it very different from what you hear on pop music? No. And that’s why no one would have any idea that it wasn’t, that it was an AI artist.

CHAKRABARTI: Yeah. Okay. And then the actual music itself, you’ve also sent some, a breakdown of the musical style. Actually Mike, he’s our actual real engineer.

Can you play a little bit more? Of Lyra Vega.

(SONG PLAYS)

Through my brain, your hologram dream. Nothing’s ever what it seems.

CHAKRABARTI: So musically, what do you hear there, Rick? I want you to, like, do the breakdown of what you hear.

BEATO: It’s an incredibly well written melody that the arrangement sounds exactly like what you would expect to hear on a well-produced pop song today. It’s really well done.

And it’s really difficult for me to say that I want to say, oh, it’s terrible. But that’s just not true. The melody is very strong on it, but it’s based on other people’s work. And other people might argue, all songwriters are influenced by artists, all their favorite artists and what they bring to the table is a combination of all these influences they’ve had, plus their own personal experiences.

So how is that different than AI? AI can create a song, including the arrangement and a perfect vocal performance in a minute and a half.

And it can continue to do that forever.

AI can create a song, including the arrangement and a perfect vocal performance in a minute and a half.

And it can continue to do that forever.

Rick Beato

CHAKRABARTI: So I just want to ask you a little bit more about, can I even call it an arrangement? I guess I can, because it’s music.

Yeah. But you’ve sent us some notes here. Like for example, in the chorus, there’s wide soaring vocal melody over shimmering synth arpeggios, and a heavy four on the floor kick releasing the tension into something euphoric, yet bittersweet. Now, is that a prompt that you put in, or is that your analysis of what the AI churned out?

BEATO: That is what AI churned out. It churned out the whole concept for it. I literally wrote three sentences.

CHAKRABARTI: And it gave you this.

BEATO: And it gave me all this, yes.

CHAKRABARTI: Wow. Okay. Yeah. Alright. Is it good? Is the song

good?

BEATO: This song is actually pretty good. I thought. It’s pretty good.

I gotta say, it’s better than a lot of the pop music. When I do, every three months I do a Spotify top 10 countdown, and I would say it’s probably better than 70% of the things I hear, 75% of the things that I hear, the melody is very strong. The arrangement, it sounds pretty good. A lot of the AI that are completely generative like this, have a sound to them.

My kids can hear these, what they call artifacts, especially in the vocal, and they immediately know it’s AI generated.

CHAKRABARTI: How, but how?

BEATO: Because there are things that are part of the reverb on the voice that have artifacts that sound almost like backwards sound effects.

They’re very faint, but they can pick them up instantly. So if I set my phone down and I played this song. My 12-year-old Layla would say, why are you listening to AI?

But these are children of a musician. And I’m gathering your kids are musically talented themselves, though.

For normal people.

BEATO: I think normal kids. If they’ve had any, a little bit of music background, they would recognize it.

CHAKRABARTI: Really?

Maybe not this song. It depends on how dense the arrangement is. The more dense it is, the harder it is to hear the artifacts.

This song, I don’t know if they would know, but the vocal is so perfect on it. Because AI never makes a mistake. Lyra is such a good singer. She never hits a bad note.

CHAKRABARTI: Lyra Vega, you heard her here first. This is her world debut On Point.

Part II

CHAKRABARTI: We brought Rick on today, not just because his YouTube channel is awesome, but because he recently put up a video.

Called, I’m sorry, but this song sucks. This new artist completely sucks. And let’s listen to the song that Rick said, it’s completely awful.

(SONG PLAYS)

CHAKRABARTI: So Rick, this is Eli Mercer.

BEATO: (LAUGHS) It’s absurd.

CHAKRABARTI: (LAUGHS) Another one of your AI creations. Okay, so I was looking at the comments in this video. Your subscribers are a fantastic bunch. Because a lot of them said, you know this is AI because no one goes to West Texas.

BEATO: (LAUGHS) Exactly.

CHAKRABARTI: They’re all trying to get out of West Texas, not being a Texan, I can’t verify that, but I thought it was a pretty good observation there.

BEATO: There were some hysterically funny comments in the comments section about it.

CHAKRABARTI: Now this again so there’s no genre specificity to what AI can do. Let me just get your broad take.

As a musician, technology has always been part of music making. Is this something though that’s just separate and apart? Is it just completely different?

BEATO: This is really different. This isn’t like when the drum machine was invented and people said, oh, it’s going to take the place of human drummers. And this is really different.

The ability to create lyrics in five seconds, and you can keep refining these things. Oh, I don’t like the second version. Can you rewrite this? This is just, there’s never been anything like this. And then the fact that you can use it in so many different ways. I had a friend who played me a song that he had written with a producer friend of his.

He wrote the lyrics. He’s a famous songwriter that’s had huge hits. So he had the lyrics and the song and the melody, and he played me the demo, and it sounded amazing. It was this country song, had pedal steel. It was a great arrangement, harmonies. I said, when did you do that? He said, oh, I did. It was Suno.

And I said, you’re kidding me? That’s all AI? And he said, yes. And there were demo studios in Nashville. There used to be a lot of them that people would go in and you’d cut three songs at a time, and it would employ people that were session players when they weren’t working on making records.

And that would basically decimate all these industries, music libraries that play bumper music that you’ll play between commercial breaks, things like that, or movies that have just the potential for disruption in the music business is incredible.

CHAKRABARTI: And we should talk about just like with LLMs like ChatGPT that people use for non-music purposes. There’s this big question of where is the, in this case, the lyrics and the songs that the AI is getting that it’s training on. And this has always been a huge issue in music in terms of are artists being adequately compensated?

BEATO: No. When the record labels, there’s three main record labels that own all the smaller ones, Sony, Warner Music, and Universal Music Group. They sued Suno and Udo who came back, I don’t know where the lawsuit is at this point, but they basically said it was fair use.

So they used all the music they wanted to train these programs and I don’t see Congress doing anything about it, frankly.

I’m not sure what you do. They have had rulings where they said that anything that’s completely generative, this would be an example of it, that’s completely generated by a platform.

You cannot copyright the song. But who’s to say that I couldn’t create this song? Get a different female vocalist, just basically replay the song and say that it was my song. Now is it my song? Does Suno own it? I asked the CEO of Suno, and he said Mikey Shulman, and he said, no, that they don’t own the song.

From what I understood. But I don’t know if that’s been decided and I don’t see, frankly, I don’t see any laws coming down about this.

CHAKRABARTI: The technologies are so far ahead of lawmaking. this huge gap that, you know, individual artists that like, they have to wade through somehow.

I just wanna go back to why you even made the Eli Mercer video. What inspired you to do that?

BEATO: I wanted people to see just how easy it is. Because I have people sending me, they’ll email me songs, Hey Rick, I wrote this new song with AI and then they send me songs and I’ll maybe type back and I’ll say, you mean AI wrote this song for you?

But I constantly am getting things like that and people really think that I guess they’re their songs. I don’t know. But I wanted to show people how simple it is to do this. You just type a couple sentences and out pops a song.

You don’t need any music training. You don’t need to learn how to play an instrument.

You don’t need to do anything except have a computer and be able to type.

CHAKRABARTI: I want you to hold that thought. Because I want to come back to it because there’s something fundamental about music making and being human that I think this is really, my personal opinion is it’s attacking that, but I know that there’s a wide variety of opinions on the good or the bad that AI can do.

But I want to play another track here. This is from the Velvet Sundown. Now that’s an –

BEATO: oh no,

CHAKRABARTI: That’s an AI band that racked up millions of listens on Spotify. And that was before people knew that it was an AI generated group. The band is a verified artist on Spotify and their song Dust on the Wind has generated more than two and a half million listens.

Here it is.

(SONG PLAYS)

Dust on the wind, boots on the ground, smoke in the sky. No peace found. …

CHAKRABARTI: Honestly, I wouldn’t have been able to tell. But since Velvet Sundown was revealed to be an AI generated group, their Spotify page has been updated. The About section of their landing page now says The Velvet Sundown is a synthetic music project guided by human creative direction and composed, voiced and visualized with the support, that’s an interesting word, of artificial intelligence.

This isn’t a trick, it’s a mirror, end quote. Okay, so with that in mind, Rick, hang on for just a second. Because I want to bring in Amanda Hoover into the conversation. She’s the senior correspondent at Business Insider in where she covers the tech industry, formerly used to work at Wired, where she covered Spotify and AI.

Amanda, welcome to On Point.

AMANDA HOOVER: Thanks so much for having me.

CHAKRABARTI: Okay, so Spotify obviously is what, maybe it’s one of the biggest music listening platforms in the world. Now, let’s start with where we are now, and then we’ll go back in time a little. What is their rule about AI generated music on the platform?

Does it have to be labeled as such?

HOOVER: Yeah. The latest I’ve seen on this is that the labeling was not strict. There’s definitely more rules around impersonating a real person, taking their likeness, violating copyright in that way. There’s rules about artificial streaming and having, like, bots listen to music, but there’s less clarity around the extent to which AI can be used to create music.

Because there’s ways that it’s used in just small ways. And there’s still a lot of human oversight and human creativity that goes into that, to fully generate songs, which potentially could violate copyright. Because they’re made from platforms that have been trained on artists’ music.

So it’s a bit of a sticky situation. And we ran into this issue with the Velvet Sundown. Because a lot of people did think that they were real.

CHAKRABARTI: Yeah. And so Rick right now, if he wanted to, he could just start Lyra Vega’s stream on Spotify or Eli Mercer’s. Put Eli Mercer’s music out on Spotify and make money from it.

BEATO: That’s not clear actually, but it does take a couple days to get the songs up there, so I couldn’t do it right now. I could do it a couple days from now. But when Velvet Sundown came out, I made a video about that and I said it before they fessed up and said it was completely generative.

Now that was another instance where my kids instantly said, why are you listening to AI? I played that song, instantly identified it, and I was reading all these articles that said, oh, you can’t, there’s no way to tell if this is AI. And I said, yes, there is. So I used another AI program to try to break the song apart, and when you break apart an AI song, you get all these weird artifacts.

So you can’t take the vocals apart and the guitars and the piano, and the drums, it just comes out with a lot of, it sounds really strange. So that’s how I knew it was AI.

And then they updated their bio.

CHAKRABARTI: Oh, okay. But Rick, again, like your ears and your family’s ears are far more sophisticated than mine.

But Amanda, do we, Rick said it’s not clear. Do we know if, again, like if I, I don’t know much about music at all. If I use this kind of similar platforms that Rick did and started making tracks, making songs and put them on Spotify, could I make money from that, Amanda?

Are people allowed to do that?

HOOVER: It’s not like just anyone can upload to Spotify at any time. But it is easier, you have to go through some certain channels. You have to. It’s not like I can just go and upload a song to Spotify today, but the bar is a lot lower for uploading content to Spotify than, say, putting a show on Netflix. So you run into this issue with the music industry where it is still much easier to get your stuff onto platforms than it is for other types of mediums.

When we think about other streaming. There was a huge issue, I believe it was two years ago about with Spotify, where people were making just like AI slop music, fully generated, just sounds essentially, not even pleasant to listen to, creating playlists of that and then having bots listen to them. And that was disrupting the way that royalties are paid out on Spotify. And that was a huge concern for artists. Because there’s only so much money, it’s divvied out into shares. Spotify, that’s against their policies.

That’s something that they say that they try to address. They kicked a lot of those songs off. But you can see that there are still ways to upload this music and potentially make money from it.

CHAKRABARTI: Yeah, because, so if you get it on the platform, you can monetize it. Okay. You started covering Spotify’s relationship with AI in 2022.

What were the issues back then that drew your attention?

HOOVER: Yeah, there were a couple of songs that were somewhere on YouTube or TikTok, but also making the way to Spotify that were in the style of, and really sounded like other artists. There was a song that went quite viral using, imitating the voices of Drake and The Weeknd.

There was a person behind this, manipulating all of this, but people really thought that there was a new song from these very famous artists, and that was raising a lot of issues for companies like Universal Music Group, who want to protect their artists and make the money off of their artists.

Whereas you have a third party imitating these people, it becomes very messy. I personally wouldn’t like it to have my voice and likeness imitated, performing work that I didn’t create. I think a lot of us would agree with that. So that was becoming the conversation just like two years ago and it did sound very convincing.

CHAKRABARTI: Okay. So Amanda, you mentioned that this track Heart on my sleeve, and that was intended to sound like Drake and the Weeknd. Again, neither of those artists contributed to the song, but we have a little bit of that, so here it is.

(SONG PLAYS)

CHAKRABARTI: Again, I’m disturbed by how I probably could not tell the difference if you hadn’t told me. But Amanda, this is definitely getting into like IP conflict territory, right? I imagine the record companies were not happy with this, or especially not happy with Spotify letting the music on the platform.

HOOVER: Yeah. They were not, they made a lot of noise about it. I believe it was taken down. But when you look at more official platforms, but once the song is out there, obviously we just heard a bit of it. Once it like proliferates on social media, it’s just everywhere.

You can never really like, fully, pull this back,

CHAKRABARTI: Rick. I said, I heard you go. What do you think?

BEATO: That’s right. Yes. Once it’s out there it’s out there forever. And the question arises, who actually owns Drake’s voice, for example, does he own it? Does the record label own it that he’s signed to it?

There’s all these ethical and do you own your own voice or does the record label who you are assigned to rights to it? And what do you do to these people that are creating these sound alikes?

CHAKRABARTI: And these are unanswered questions so far as, as far as I can tell Amanda?

HOOVER: Yeah, I think that there hasn’t really been a very definitive answer. I think that it’s really interesting from this. I think what this showed us was something really interesting culturally, because people liked the song. People were excited about it, it went viral and it went more viral once people knew that it was AI generated, but before that, people liked it. So I think it really raised questions for the first time. People weren’t very familiar with how advanced these tools had become. A lot of people had never really heard what they could do and it was really raising questions for people of does it matter if I like a song that’s made by AI?

Is that taking out some of the human artist interaction. What does it say about me as a listener? That  I want to hear this. What does it mean for artists? I think all of those questions are also pretty unsettled.

CHAKRABARTI: Amanda Hoover, senior correspondent at Business Insider, where she covers the tech industry.

Thank you so much for joining us today.

HOOVER: Thank you.

Part III

CHAKRABARTI: I want to talk with you now, Rick, a little bit about, we started off by talking how about how easy it is to just generate like brand new fake artists and new songs just from a couple of prompts. But a step back from that is how musicians are using AI already in the process of making music.

So not just generating fake artists, but actually using AI as a tool. So here’s a little example. During rap’s recent major feud. Amanda talked about that a little bit, between Drake and Kendrick Lamar.

Producer Metro Boomin released a track called BBL DRIZZY, BPM 150. The track sampled the song BBL DRIZZY by Comedian Willonius Hatcher, known as King Willonius online, and here’s some of the track.

(SONG PLAYS)

CHAKRABARTI: So Billboard calls that the first song released by a major producer that relied on an AI generated song for a sample. Okay? So that’s the key part. Now, that track, as I said, was based on BBL DRIZZY by Willonius Hatcher. Known as King Willonius online.

And here’s the original completely AI generated track.

(SONG PLAYS)

CHAKRABARTI: So Rick, is there room for AI? I guess it’s already happening. What do you think about using AI as a tool in this way in music production?

BEATO: I think it’s great, honestly. And people have been doing it now for, geez, probably two years or so. I know writers that every writing session, ChatGPT or Claude or Gemini, any of the AI LLMs, they use it for lyric writing.

I know writers that every writing session, ChatGPT or Claude or Gemini, any of the AI LLMs, they use it for lyric writing.

Rick Beato

That’s incredibly common. And now these platforms like Suno and Udo, they use ’em for idea generation. Because it can generate melodic ideas, harmonic, meaning chord progressions, and lyric ideas. It can help you with writer’s block or exploring things that are outside your comfort zone. You can do rapid prototyping.

It can save you time, like I was saying about creating a demo of a song. If you have it already written, but you don’t have the players to play it, and you may have all the chords and the melody and the lyrics, it’ll create the demo for you. So it’s incredibly helpful actually.

CHAKRABARTI: Do you use it?

BEATO: I use it to make these videos, but I’ll use AI … to organize thoughts.

For example, if I have a video that I’m making on a particular topic, and I’ll just type in a bunch of random ideas and I’ll see what comes up. And sometimes it will organize them, but I don’t use it for idea generation, because I make videos about things I’m interested in.

And which sometimes is AI. Did I rely on AI to make this video? I used it and I did it more as an informational tool to show people multiple platforms. Because a lot of people watch my channel.

Yeah.

And if they are not familiar, I think most people are familiar with things like ChatGPT.

They know that term they’ve used it. But some of these other ones like Claude and Gemini and Grok, they may not know those. So I try to include as many as possible in there. Yeah.

CHAKRABARTI: Idea generation. This is really interesting to me because in all of the AI shows that we’ve been doing, there’s an emerging theme about the concern that like a lot of use of AI produces a kind of homogeneity.

We just did this show about AI and writing, and we talked to an MIT researcher who did a pretty significant experiment with students asking them to write using different tools. The students that used AI exclusively, their writing, even though it was different, students came out pretty much the same.

It was very similar. And to me, this is one of like the big concerns with music, especially because it is, like, I guess what, humans, birds and whales or sea mammals, those are the only ones I can think of that make music. And it’s such a human thing that we started off talking about my obsession with your video about Kiss from a Rose.

That to me is like a perfect example of, I don’t think AI could have ever created a song like that. It is so unique and so beautiful. And so just the second you hear it, you think, oh my God, how did Seal do this? Do you ever have concerns that like, that kind of unique musical moment could be threatened as more and more artists start using AI?

BEATO: I think some of the areas that people will move towards is more improvisation and live performance and trying to write things that haven’t been written before. Because there’s plenty of music that’s not been written. A lot of the platforms or genres like pop music, EDM or dance music, country music, things like that will use a specific kind of harmonic and melodic template in order just to be within that particular genre. Any of these particular genres are easier to mimic than something that may be a progressive rock song. So maybe new styles of music will emerge that haven’t been trained by these AI platforms.

CHAKRABARTI: Okay.

So speaking of pop music. Is the AI so good at replicating or mimicking pop music? Because pop music is so bad.

BEATO: (LAUGHS)

CHAKRABARTI: No. I’m talking about, it is thoroughly, it’s always been, but now more than ever a commercial endeavor, right? Like a song is defined, in what? The first two seconds. And it’s always written by a team that isn’t necessarily trying to write the best song, but they’re trying to write the catchiest song so that they can market it. So I’m just wondering if AI is just a natural extension of that in music.

BEATO: Yeah, I would say it’s also about the people’s willingness to accept this music. When people hear these things, when I play, when I just heard the Eli Mercer song, I hadn’t heard it since I made the video last month, and when you played it there, I thought, oh, Eli Mercer songs. That’s not too bad. The chord, the lyrics need a little work, but I kind of like his voice.

So people will readily accept AI music. They already do. They don’t even know they’re hearing it. Especially on things like chill music or atmospheric music that you’ll find on playlist on Spotify. Now they say that they’re not using any AI music, but they can say that. But there’s —

People will readily accept AI music. They already do. They don’t even know they’re hearing it.

Rick Beato

CHAKRABARTI: So we’re hearing it and we don’t even know it.

BEATO: Oh, yeah.

CHAKRABARTI: Do you think Spotify should actually put, I don’t know, like an audio label at the top saying this is AI generated?

BEATO: Yeah, I think they should if it’s completely AI generated, yes. But will they do it? I don’t know. Do they need to do it? They don’t need to do anything really.

They own so much of the market share. But are people going to want to go and listen to their favorite artists that they can go see in person? Yes.

And that’s a big part of it. I can’t go see Eli Mercer or Lyra Vega. I could see a Lyra Vega. Somebody could change their name to Lyra Vega and do these songs, but —

CHAKRABARTI: Yeah. Listen, I just want to introduce another quick voice here because we’re trying to think not just about the future of music itself, but for the people who make the music. Jonathan Wyner is a professor of music production and engineering at Berkeley College of Music in Boston, Massachusetts.

And he said he doesn’t actually fully fear for his students’ ability to get jobs in the music industry in the future. Because he told us that learning to use evolving technologies has always been part of the work.

JONATHAN WYNER: Ultimately that’s going to support the kind of adaptation that people need to make in order to find their way to jobs making audio, making music.

I think without that, you’re going to ultimately find yourself in a world where other people are using this technology. And we’re right at the leading edge of all of this. Generative music is not, it’s less than 1% of the streams on Spotify, so it is not a phenomenon right now that’s threatening anybody’s job in that way.

But long term, you could imagine that it will increase as a phenomenon and people need to know how to use the tools.

CHAKRABARTI: Jonathan Wyner says, though, that AI does have the potential to affect how consumers use streaming services and how musicians then are able to find their audiences.

WYNER: The streams being flooded by the billion tracks that could be generated by AI compared to the tens of millions generated by humans … you start to undermine whatever income stream there is from streaming. Do I worry about that? You bet I do.

CHAKRABARTI: And he says that much like writing, people are drawn to the act of creating music. So there will always be something deeply human about human made music, he says, because putting in the hard work will not be deterred by AI’s proliferation.

WYNER: We’re talking about being creative and there’s nothing very creative about pushing a button that says song and getting a song back. Most people who get involved in making music and stick with it are, in my experience, interested in improving, getting better, which leads them to think harder about what they’re doing and how they’re doing it and what they’re hearing.

CHAKRABARTI: So that’s Jonathan Wyner, professor of music production and Engineering at Berkeley College of Music. Rick, your quick thoughts about that.

BEATO: I totally agree. Everything he said. People play music because they like to do it, they enjoy it. They enjoy, from what I’ve found, they enjoy getting better at it and the process of it.

And you can go out and play for audiences. So I don’t see AI right now taking over. Yes, it can make billions of songs, things like that. But, to his point, it’s a small part of the market share right now on these platforms like Spotify, and I’m not worried about it right now.

Personally.

CHAKRABARTI: Right now. Exactly.

BEATO: Yeah. But six months from now, yes.

CHAKRABARTI: Who knows. Yeah. Because the incentives, the business incentives to me seem more aligned with AI than for the platforms and supporting human made music, right? Because AI is basically instantaneous and free.

And if it keeps people on the platform, that’s really what Spotify wants. Can you imagine any kind of other kinds of reverse incentives to bring the platforms back to actual, like human musicians?

BEATO: I think it’s just gonna force musicians to be more creative in their presentation, their live presentations or online videos.

Having YouTube channels, or TikTok or Instagram, things like that. And creating a connection with their audience, in addition to their music, is that they’re forced to put the face to the music, even put their face to the music even more.

CHAKRABARTI: Yeah. But like you said, you can’t go see Lyra Vega live.

BEATO: Yes.

CHAKRABARTI: We’ve only got two minutes left, Rick, and I’m suddenly seized by, even though I see the potential of AI, I am also seized by this fit of nostalgia. So I want to play something here that involves one human. This is 1982. Michael Jackson sang his own background vocals in one of the bestselling albums of all time.

Here’s what that sounded like, the background vocals, the construction of it, track by track.

(SONG PLAYS)

CHAKRABARTI: Okay, so this is from Robert Komaniecki, a professor at the University of British Columbia School of Music. And he did use AI to actually separate out those tracks, we should say, but bring it together and you get this.

(SONG PLAYS)

CHAKRABARTI: Still one of the greatest songs of all time.  What makes that great, Rick?

BEATO: It’s human. It’s incredible. It’s so well done. All those harmony parts, so well thought of and so well executed. Michael Jackson had incredibly good pitch, just an amazing singer. And just the tuning of that, this is before autotune, is so precise.

It’s really remarkable.

The first draft of this transcript was created by Descript, an AI transcription tool. An On Point producer then thoroughly reviewed, corrected, and reformatted the transcript before publication. The use of this AI tool creates the capacity to provide these transcripts.