AI music is here

♪ We are Charlie Kirk, we carry the flame ♪ ♪ We'll fight for the gospel ♪ ♪ We'll honor his name ♪ A.I. Music is everywhere. It's slopping around your algorithm and your platforms. One of them, dessert, says 50,000 A.I. generated tracks are being uploaded every day. Spotify is declining to comment LOL. And music is also charting, breaking rusts, walk my walk, top spotifies viral 50 songs in the US. Zaniamone debuted on the Billboard charts. And yet, many people cannot tell the difference between AI music and music music. And if, to paraphrase Taylor Swift, that makes you want to die, and the sun, do we have a show for you? Coming up on today, explained, "Should music lovers take AI music seriously?" Support for this show comes from Vanta. Vanta uses AI in automation to get you compliant fast, simplify your audit process, and unblock deals, so you can prove to customers that you take security seriously. You can think of Vanta as you're always on AI-powered security expert, who scales with you. That's why top startups like Cursor, Linear, and Replit use Vanta to get and stay secure. Get started at Vanta.com/vox. That's V-A-N-T-A.com/vox. Vanta.com/vox. Every year, hundreds of thousands of people from all over the world flock to Las Vegas for the Consumer Electronics Show. And they spend a week trying to sell each other on the weirdest gadgets you've ever seen in your entire life. This week, on The Vergecast, we're talking all about everything happening at CES, from the TVs, to the AI gadgets, to the humanoid robots that everybody is hoping might someday do your laundry and wash your dishes. All that, and much more, on The Vergecast, wherever you get podcasts. Ian Kreitzberg is the AI correspondent for Puck. He writes a newsletter called Hidden Layer. And this year, like many of us, Ian encountered Velvet Sundown. Velvet Sundown. They're a band that I guess they kind of, they got really popular. I say they're a band. They're an AI music project. Is perhaps more accurate. They got popular a few months ago around the summertime. And they do a kind of iteration of sort of 1970s inspired classic rock. That, if you're a fan of 1970s classic rock, it sounds pretty derivative. And they had a bunch of tracks exploded in popularity. Their number one song is called Dust On The Wind. Dust on the wind, boots on the ground, smoking the sky, no peace found. Rivers run red. Reminiscent. A little? Of a little something. From Who Is It Kansas? Yes. The way Spotify works, right, a lot of music discoveries through Spotify's playlists. And so they got music in the playlists. And they were one of these examples, as I was mentioning earlier, right, when there's no labels about this. They weren't talking about it. They were kind of gradually found out by people and listeners who slowly put it together. There were a couple of red flags. For instance, they released a lot of albums out of nowhere and in very quick succession. In the span of a couple weeks, three or four records out, which is a pace that raises eyebrows. And so people started putting it together. And eventually they said, you know, hey, we are an AI music project. They got a lot of press. A lot of people were checking them out. And so that they were an interesting example of, you know, the idea that you can kind of do people into listening to fully AI-generated music. And it's not immediately apparent that this was not produced by people. We've spoken to people for this show who really love music. And some of them, you're not going to be shocked to hear this. Some of them are really kind of appalled by the idea that people are listening to AI-generated music. It has no heart. It has no soul. There's no real connection between the quote-unquote artist and the listener because there is no real artist. It's artificial intelligence. But here's what I wonder. Your average listener is a person who is not a super music nerd. Your average listener is a person who's on a platform. And stuff maybe is getting fed to them. And they're like, oh, I like this. I don't like this. What has your reporting told you about how the average listener thinks about AI music? My answer is not one that will make those music nerds, which, you know, I would include myself as one of them. But very, very happy. I think the reality is that a lot of people just don't care. When, you know, in the earlier days, this AI music didn't sound good. It didn't sound right. It was difficult to produce full tracks that were convincing. When you surpass that point and on a casual listen, it sounds like a normal track. People don't care. The fascinating thing, though, is that when it's labeled as being AI-generated, people do tend to care. Oh. Right? That was my reaction as well. So what, so what, so they avoid it? They don't click on it. What do they do? So they've done studies of art. So they'll show images, right? Where if you label an image as being AI-generated, people tend not to like it as much, as they like an image that is labeled as being created by a person. If we talk about the major players in the AI music space, what I'm coming to learn is we are not talking about individual people making music and then generating it through AI, we're talking about platforms that let you make music using AI. Who are the major players in this space? There's really two major platforms here. It's Suno and UDO. And they both do essentially the same thing, which is exactly what you just said. Platforms that allow you to generate music through text prompts. Okay, so I would write in, I want to stirring early morning gospel song in the style of "Rise and Shine and Give God the Glory." And then it makes me want, just like that. You would get a stirring gospel song, yeah. Wow, okay. So let's talk about the reaction from major record labels. They see this thing on the horizon. I assume they know it is likely training on their artists and they do what exactly? Well, they did a couple things and we're talking at an interesting time when we can kind of look back and see the whole arc of the reaction. And so the first thing they did was they sued. A legal firestorm. It's brewing today over this major clash between record companies and AI music services. The Recording Industry Association of America filing copyright infringement lawsuits today against two separate companies. The RIAA announced that major record labels are suing two of the leading AI music companies alleging a massive copyright infringement and is maybe trying to shut them down. Universal and Warner, they filed lawsuits for copyright infringement. These lawsuits were focused on two factors. The first is the inputs and the second is the outputs. And this is where the kind of idea of fair use comes in. And so in that first, the inputs factor, right, the loose argument was basically something along the lines of you illegally took my music, my content, my data, and you used it to create this model that you're generating revenue from. You didn't ask me if you could do that. You didn't pay me for it. And so we're suing you. Then there was the output side, which is, okay, regardless of the legality of what you did in training the models, your model produces music that is often, and then they cited specific cases, quite literally identical copies of music in our catalogs. That was a more convincing argument, but both of these suits were settled recently. And now you have partnerships, Universal partnered with UDO and Warner partnered with Suno and Universal announced a partnership with NVIDIA to "transform music experience" for their fans with NVIDIA's AI. So you see this shift where they kind of seemingly use their litigation, and these are like major companies, right, UMG and Warner, to involve themselves in these companies that were certainly, it's not hard to perceive them as threats if you're a record label, as a means of what seems to me to be, you know, hedging their business in the future. Okay, so I'm thinking about this from the perspective of the record label. I want to make money off of music, and I'm aware that AI is out there, and I have to live with it, so I'm starting to get used to it. But I also know, based on what you told me, that people don't like knowing that they're listening to AI music. They feel like there's something uncool about it. So if I'm a record label, and I say, I want to put out AI music, because I see it as the future, how do I sell that to people? A, they don't know. What the hedge is essentially looking like from where I sit, is there's a non-zero chance that in some manner, AI becomes the primary way in which people consume music related content, or just all content. And if that does happen, these companies want to make sure that they're involved in that, that they have partnerships with these companies, or that they have equity in these companies, right, that their business is encrypled if the way people consume content changes dramatically, and we never go back, right. Now, I don't know that that would happen, and the idea of, you know, can UMG sell an AI artist, for example, can they sell straight up AI generated tracks? That would be a hard sell. There's a lot of artists on these labels that are not a fan of what's happening here. These labels don't want to piss off their major artists, because that's their current business model, and making sure you have the fandom of Taylor Swift is important to not push away. I think that kind of idea of how can we allow users or consumers, people and audience, to remix something they already like, to live in it in a different way, to somehow customize it to them. I think that's what the record labels are looking at as a potential way to not alienate their current artists, and not alienate and destroy their current business as, you know, record label of people, while opening themselves up to additional opportunity in that kind of new experience, whatever that music experience looks like, while hedging themselves, you know, in case that new experience becomes the only experience, with the only major experience. That was Ian Kreitsberg of Pocky Reports on AI, coming up a tale as old as time, a dude, experiments on himself. So, "Poor for the Show" comes from Quince. New Year, New Year, sounds unrealistic. Why not start with New Year New Wardrobe instead? For that, there's Quince. Quince makes quality wardrobe staples that are stylish and affordable, from soft Mongolian cashmere sweaters that feel like designer pieces without the markup, to 100% silk tops and skirts, for easy dressing up. Their wardrobe essentials are crafted to last season after season, and according to Quince, each piece is made with premium materials then priced far below what other luxury brands offer. Our own Nisha Chital has gotten some Quince. I got the organic cotton boyfriend crew neck sweater. I'd seen it around on social media a lot, so it's a great, just classic, oversized sweater. I got it in a cream color, and I think it'll work with a lot of different things and be easy to build an outfit with. Refresh your wardrobe with Quince. Don't wait. Go to Quince.com/explain for free shipping on your order and 365-day returns. Now available in Canada too. That's q-u-i-n-c-e.com/explain. To get free shipping and 365-day returns, q-u-i-n-c-e.com/explain. Support for today explained comes from Vanta. Vanta says if you run a business, you know how important it is to keep your customers trust. Frankly, since Vanta maintaining that trust can make or break your business, makes sense. But the more your business grows, adds Vanta, the more complex your security and compliance tools can get. And left unchecked Vanta adds that can turn into business chaos, and furthermore, chaos isn't exactly something that customers trust. Nor is it a great security strategy. That's where Vanta comes into Vanta. Vanta says you can think of them as you're always on AI-powered security expert who can scale with you. How do they do that? She asks. Vanta says they can automate compliance, continuously monitor your control, and let you look at your entire compliance and risk ecosystem from one place so you can see the whole picture. Perhaps you're a fast-growing startup like cursor, perhaps you're an enterprise like something called snowflake. Vanta says they can fit into your existing workflows. You can stop spending so much time worrying about compliance and instead focus on your customers. You can get started at vanta.com/explain. That's v-a-n-t-a.com/explain. Vanta.com/explain. Support for today's playing comes from Bombas. Perhaps you want to get in shape this year. Bombas wants to tell you about the all-new Bombas sports socks engineered with sport-specific comfort for running, golf, hiking, skiing, snowboarding, and all sport. Meanwhile, for the loungers among us, Bombas has non-sport footwear available. But Bombas doesn't just offer sport and non-sport socks. They also offer super soft base layers that they claim will have you rethinking your whole wardrobe, underwear, t-shirts, flexible, breathable, buttery, smooth premium every day. Go twos, they say you won't want to leave the house without. Here's Nisha Chital. I've been wearing Bombas for several years now. I have several pairs. My whole family loves to wear Bombas. I have several pairs of Bombas ankle socks, and I have some no-show socks as well that are great for things like loafers and ballet flats. For every item you purchase, Bombas says an essential clothing item is donated to someone facing housing insecurity. One purchased, one donated, over 150 million donations. And counting, I'm told, you can go to Bombas.com/explained and use code explain for 20% off your first purchase. That's B-O-M-B-A-S.com/explained, code explained at checkout. This is an artificial intelligence version of Drake and Duelist's name, told today. Explain. Noelle King here with Denise Bishard. He's the senior tech reporter at Scientific American. I reached Denise a few weeks into an experiment that he's been running in which he is only listening to AI music that he generates himself. Why? Because he's Denise. What? The reason I did it was because I wrote an article about AI generated music, and in the article I was trying to decide can AI music really appeal to us, can it be important to us, can it be something that is part of our lives? And I went back and looked for a song that had always been there. It had been there for a long time, which is for what it's worth by Buffalo Springfield. For what it's worth as a protest song written in 1966, in response to the sunset strip curfew riots in Los Angeles. And this is something I learned to play on the guitar from my stepfather when I was a teenager. And I tried to have Suno create a protest song for me to see how I would react to it if I would have a similar emotional resonance or whatnot. I asked for a song that was folk, warm, had male lead vocals with Ernest Tone, steady mid-tempo groove, acoustic, and had a vintage texture with a subtle tape hiss. And it took that and developed that, and something that was quite different from what I had actually asked for. But it still had that feeling of that time, still had lyrics that had a bit of a protest song quality. And if I had heard that song in a coffee shop or restaurant, I wouldn't have known it was AI. It was very well made. But what I realized was that there was just no story for me. There was no attachment to that song. I found that there wasn't that level of human connection to the AI song. And my conclusion was kind of, well, you know, this music doesn't have a story with which we connect. And then after I published that, I thought, hmm, well, maybe it's just because I haven't given it a chance. You know, maybe I haven't really explored this question fully. So I thought, I might take a month and just listen to AI music and nothing else. And I will come up with a prompt and I'll plug it in and each prompt makes two songs. And I'll try to be as creative as possible. And I'll usually plug it in three or three times and vary it. Add different kinds of instruments with it or different kind of vocals with it. And just plug a bunch of those in. For example, I asked at one point to make this song that was a mix of rap and bluegrass. And it made a song called My Existence. And I listened to that one a few times. I was kind of intrigued by how it made that one that made me laugh was a song called organ trafficking. I'd asked for contemporary rap song with female vocals, playful, ironic lyrics. And it came up with this song where organ trafficking is kind of the central metaphor. I was pretty surprised. I was thinking, okay, I didn't really expect AI to do that. But then, you know, there are also moments in the evening when I kind of want to wind down. And I will look at the prompts and I'll think, okay, give me a song with literary lyrics, playful metaphors, soft female vocals, and acoustic guitar. And then it will give me something like shadows of the sun. But it tore right through and drank me up on chasing shadows of the sun, running in circles till I come undone. I'm a flame that drowns in a wave that burns every wrong turn teaches the heart. And I have also done to play the things where, you know, I've asked it for Boston Nova and it makes it with rap. And it gave me a song in Portuguese called Entry of the Sailor Fogo. I think one of the things I've realized is that a lot of the music I listen to that is mainstream, that is. I would consider kind of heavily processed music, music that's designed to have a large market. It doesn't feel very personal to me anyway. So I realized that in that particular context, it didn't feel very different. A lot of the time. Do you think if someone had handed you a playlist and ten songs, five or AI, five or not, do you think you'd be able to tell the difference? No, I don't think so. Wow. And what is that? What does that tell you? I mean, it tells me that AI is getting very good. It's certainly telling me that. One thing I noticed during this process was that a lot of the AI music that is popular, that people listen to on Spotify. That has millions of listeners. It often is creating songs that are very soulful, very gritty. You know, it's like "Is there any Mone" or "Solomon Ray" or "Cane Walkers" don't tread on me. "Cane Walkers" is not a person, it's an AI avatar, right? Or breaking rests, living on borrowed time. And if it's cars, they're proof I'm alive, we're all just living on borrowed time. Those songs all feel just really authentic. You know, if I were to sit down in a bar or somewhere and someone picked up a guitar and sat, I would expect some gritty local musician to get up and sit on the stool and sing one of those songs. And think, yeah, this is really authentic. This person really suffered through these things and felt these things. That's how they come across. And I think that AI tends to work best when it just leans into that authenticity because it kind of helps overcome the cognitive dissonance that we're thinking this isn't really a deeply felt song. And it moves away from mainstream human-generated music, human-made music, which is often very heavily designed to be a summer hit or to go viral in some way. And it doesn't often have that level of authenticity, that feel of authenticity. And I think when AI replicates that, we're more aware of it being superficial or artificial because there's already an element of artificiality there. Do you think when your experiment is done, you're going to keep making a music? I think I probably will. You love the power. And I think what has surprised me with it is now I'll be walking somewhere and I'll think, what if I were to ask it to combine these styles or put a banjo with a hip hop track and add this kind of old clothes? I would say now I'm at the point where I don't worry about the connection to the human, like I did in the beginning and the beginning, I was like, who's this person? You know when you're reading a book and you're halfway through the book and you think, what human mind did this book come out of? And you turn the book over and you look at see who the author was and you Google them and you're like, how in the world did they think of this? And I just had that impulse so often the beginning to want to know, who felt this? Who thought this? I just would have cognitive dissonance. This is a machine. This machine did not fall in love. This machine did not suffer these experiences. This machine did not wake up at 2 in the morning and write this song, just needing to express itself. I was actually really bothering me. I was kind of blocking from being able to enjoy the song and I thought, well, if somebody created an AI avatar and gave it a personality and there were a fictional character that existed in the metaverse and that AI avatar was a song maker and that was singing this song would that make it easier and weirdly it would make it a little easier. Does doing this experiment and seeing how you're reacting to this music? Does this change how you think about AI at all? I think my conclusion from this is that in 10 or 15 or 20 years, there are going to be a lot of teenagers who look at the discussions we're having right now and go, what are these people talking about? This is totally normal. Why would anybody feel so conflicted about this? I think we're going to adapt to it pretty quickly. That is my gut feeling. There are a lot of big questions around the creators and protecting artists and what it means to be an artist. There are a lot of questions that are going to come out of this and I really hope that artists are as protected as possible and remunerated properly. But I think this is going to fit into our lives a lot more smoothly than I think we're realizing at the moment. The rest of our team includes Eddie Maul and Katie Maul's Bryan, Peter Balloon and Rose and Patrick Boyd, Kelly Westinger, Ariana and Sputu, David Dottasho, Dustin DeSoto, Estet Hunden and Sean Remaze were on. Our supervising team includes Avi Shillard, Ciamena Alcide and Jolie Myers. We use music by Brakemaster Syllinder. Credits there by Suno and you can read Denise Bischard's forthcoming piece about his experiment in Scientific American Soon. Personal question while we're here, did you buy any chance to sign a prenup when you got married? If so, can I talk to you about why? Call me at 844-453-4448 and leave me a message I want to make you famous. I'm kidding, but I might in fact want to put you on today, explain. Today, explain is distributed by WNYC in the show as a part of the Vox Media Podcast Network. For more award-winning podcasts, you can visit podcasts.voxmedia.com. Listen, sans ads by signing up at vox.com/members. I'm Noelle King. It's today, explain. [BLANK_AUDIO]