When it comes to the right and wrong of using AI in creative work, it's safe to say we're still figuring it out. That goes double for music, where AI has cultivated more than its share of lawsuits and suspicion from artists.
From the audience perspective, AI is arguably changing things even more. Recommendation algorithms have the power to influence our tastes, but they also can put us in bubbles. At the same time, AI-cloned voices paired with auto-generated lyrics could end up changing our expectations of what's good entirely — we may not care whether it's Drake or fake.
For some thoughtful opinions on all this, I looked to authorities in both entertainment and AI. At the recent Ai4 conference in Las Vegas, I sat down with none other than comedian and actor Cedric the Entertainer as well as Alex Elias, CEO of AI recommendation engine Qloo. Our conversation got into the potential benefits and drawbacks of AI in music and show business.
Despite concerns about what AI could mean for artists' legacies (Tupac does come up), both Cedric and Alex expressed cautious optimism about the future of AI in music. Ultimately I came away convinced that it comes down to finding the right balance — harnessing AI's potential while preserving the human element.
Learn how to use AI in journalism, PR, and marketing. Sign up for one of The Media Copilot’s 100% LIVE AI training sessions today!
It was a fun and thoughtful conversation, and be sure to listen to hear Alex's story about how over-producing a saxophone solo to perfection totally ruined it. And remember, if you enjoyed this podcast, it would be great if you could follow the show on Substack, Apple Podcasts, Spotify, or any other podcast app, really. Also, I’d appreciate it if you’d leave a rating or review — it really does help the show. And if you’re on YouTube, please like the video and subscribe to the channel 🔔
Pete Pachal: So, yeah, okay. So, I mean, for you, especially, like, tell me about your relationship with AI. And like, you know, you mentioned a bit of it over in the talk there that it sort of probably predates what currently we think of as AI, yeah, but in terms of how you've been thinking about it even before, was like a thing in its current incarnation. How have you had a relationship?
Cedric the Entertainer: Yeah, you know, for me, I always thought of AI, of course, as the ability to take what it is that you already know and then have someone enhance it, basically. So, you know, as a creator. As a creator, I was a person, you know, you write jokes, you write you write movies, you write TV shows, and they just kind of come out of your head, like all the all the time. But the speed in which I could do that was always affected by, again, in a way, the Add of thinking of a whole another idea while I'm doing one. So with AI, you can literally just say what it is that you're saying, shape it, have the idea really formulated into something that's tangible and usable, and then go back and enhance and work it better. So that's been my biggest use of for it. And I write songs sometimes. So I use it for everything, man. I'm like, you know, and I'm still very old school. I still have a bunch of notebooks as well, like
PP: Yeah, I know, you gotta, gotta get back to the analog
Cedric: You gotta do it.
PP: So what were your first impressions of, like, chat, GPT when it came out? Like, not just first using it, but then also, like, it must have dawned on you that things were starting to change.
Cedric: Yeah, I mean, I thought that it was, it was really sharp when, you know, when someone first told me about it, I didn't even believe it was true. I was like, I don't think they can do that. They was like, No, it will write it in your voice, or it will be the way you speak, in a way. And I was, no, they can do that, right? I'm sure they can do this. But I know my son did it. Did and I did it. I pitched him. I told him, like, oh, do this show idea, and let me see what it, what it would do in the in the in, you know, in the voice of say, to be entertainer. And it was it. It was dead on. I mean, you know, not, not with the perfect high quality restaurant, quality jokes and I would write, but again, good, good material, funny thought process. How we got to it, I was quite surprised by it. Nice.
PP: So tell me a little bit about Qloo and like, what you guys do, it seems like, obviously it's very AI in that you're kind of taking a lot of what's happening right and then figuring out tastes and recommendations.
Alex Elias: Yeah, what is Qloo? I mean, if you caught the talk, we chatted a little bit about it, but we the genesis of it was really generating better models and better predictions around sort of taste categories of esthetic preferences, in essence, so you know, the media Nexus, music, film, television, podcasts, the transactional, Geo, locational categories like travel and restaurants, and really bringing it into a meta model, kind of a holistic model that could generate some of the more unique aspects of it are being able to generate kind of cross category inference. So taking, you know, signals in one category and generating predictions and others. And as it turns out, with the privacy landscape today, that's a hugely powerful kind of way to deliver value to companies without using any identity data, because what you can do is, you know, if you're a company that has data points x, y and z, you can essentially enrich that in a totally ethical, anonymous way.
So you think about, you know, everything from, we work with a large ticketing company, one of the largest in the world, and they have very sparse data points per capita, you know. So people only tend to buy tickets ever, ever so often. But at the same time, they have this ephemeral good events come and go. You miss the boat. You're done. You know, you get you have to, there's shelf life of a ticket is pretty short. So what Qloo can do in contacts like that? You could take a very sparse amount of information. So just the fact that someone bought a show to see, you know, Cedric at the at the Cosmo here, where you come, coming back, so if someone bought that, what does that suggest about their probable taste in music and sporting events and so on, and then vice versa, if someone just bought a single music artist ticket, you know what predictions includes optimized, it could generate multi millisecond inference across all those categories.
And what's been interesting is with the LLM space. So language now is kind of just the greatest interface over data. You know, it's but it's ultimately, it's ultimately an interface, and it depends on data. And so some of the, some of the biggest areas of kind of commercial address ability we now see is helping to guide L. LMS for commercial purposes. So if you think about you know that same that those same sorts of examples where you have a ticketing company now that wants to, for instance, generate itineraries for travel and have an have it be a language interface Qloo in those in those contexts now is being heavily used as an oracle of taste, so to help guide what the LLM actually narrates, because we can generate, again, this very structured inference, we have these very delineated areas of knowledge. So yeah, that was but what's exciting about it is it didn't really start.
I mean, when Cedric first invested in the company, it was just an idea about making better recommendations. It was an idea that current recommendation models were broken. You had the kind of UGC space you had, you know, the popularity contest of a good reads, a Yelp, a Trip Advisor, an IMDB, and then you had algorithms that were sort of very siloed and kind of opaque and part and parcel to a Spotify or a Pandora. And so we wanted to kind of do better in that regard. So it was really kind of almost an academic premise at first, but we certainly couldn't have predicted the fact that 12 years later, we'd be powering, you know, llms or powering certain commercial integrations to the tune of like, 3 billion network calls a day into our service.
So the in the main way people hook into Qloo through APIs, application programming interfaces. So you can kind of build on top of it. You can build other software on top of it. It's real time. And then we acquired this company that we talked about, you know, on stage called Taste dive, which kind of puts us back at the roots of when we started the company. So taste dive is a pure play recommendation engine. You know, a lot of the traffic is you type in things like movies like, anything like, if I were to type in, you know, on Google right now, movies like Rushmore, like, show me movies like Rushmore, or anything else, it tends it's almost nine times out of 10 the number one result. And then you'll have, you know, you have the ability to not only get kind of recommendations intra domain, but even cross domain. And this is your okay, this is taste type, yeah. Taste type. So, yeah. So 10,000 plus individuals on that platform have kind of opined on Rush more. And then he can sort of, you know, it's one of the cleanest, it's totally ad free, one of the cleanest ways to kind of discover the implications of your tastes, and so on. And so, yeah, it kind of puts us back at our roots.
PP: Yeah, I gotta say I kind of like, when the guys asked this question, I was had a lot of it resonated a bit with me, because I feel like recommendations bring up a lot of questions. Like, I think most folks have obviously encountered a recommendation engine account with something. I watch a lot of YouTube, and I feel like one of the big problems with the things that recommends me is it doesn't really, there's no real breakout of like, well, what about and it might be a little different, because I wasn't asking specifically, like, in the taste thing you just showed me, it's like, I specifically am asking deliberately. I want things like this, but for a portal like YouTube, or a portal like Netflix, you know, I feel like, is there? What's the opportunity to break out of that pattern and give you something that is unexpected? Because, frankly, I go to the YouTube home page from just like, I'm just like, Okay, I've had enough of all this crap. Can give me something completely different, right? Right?
AE: Absolutely, yeah, I think promoting discovery. And those are things that could actually be explicitly baked into the the algorithms. And it's also something that you know Qloo from an enterprise perspective, we have a parameter in it, where you can actually adjust for novelty and discover the wild card factor or something, yeah, to help promote diversity of results, the heterogeneity of it, we worked with a large in flight entertainment company that was programming for various flights, and they specifically wanted more heterogeneous results For for certain, certain legs, so that it wasn't, you know, it didn't just curtail the kind of what, what's being put in front of people. So, so, yeah, there's definitely, that's a you definitely want to break out of that bubble, right.
PP: At times, at times. Yeah. So I wanted to ask you about AI and music creation, if you don't mind. So the talk that preceded yours was actually a guy that was talking about his service, that was about cloning artist voices and creating new songs, but in a legit kind of way, right, in a licensed kind of way. This is obviously a huge issue that's going on right now. What is your like? And I feel like it fuels the Friend or Foe sort of framing that I think in a lot of people's minds. Just, I'd love your thoughts on that. Where do you what do you think of it? What do you think is going to go?
Cedric: You know, it was, you know, that one thing I thought was this big Drake kindred challenge that happened in the rap business, and Drake. Did a what I thought was Uniquely Creative. However, it was about it backfired on him as an artist. He used his ability to write, but used Tupac, and so someone who had passed voice on the record, and I think he used Tupac and Snoop to write a West Coast dis record, and it was creative, because as a writer, he was able to write a lyric and then use someone else's AI. And when you first heard it, I was like, That's crazy. That's super talented, but the pushback was immediate, like, it's fake, it's false. It's not Tupac. It's not don't do that, right? Don't be Tupac using in his voices, and not be him. And I can, and that's the first time I realized, like, yeah, as much as I love Marvin Gaye, I don't think I want somebody to write a Marvin Gaye song and then put Marvin Gaye on the voice now, like, let, let the music and his art and him as this creative vessel and being, if it's over, it's over, that's what it was. So the uniqueness, uniqueness of his voice and his tone and those things I really like, you know, yes, would it be great to hear some new music nowadays. Prince Michael Jackson, sure, but yeah, you know, it's I, and I had to agree with, yeah, let's, let's, let's stop the line somewhere.
AE: So there's an interesting kind of a corollary with there's a pretty big tenor sax player plays a lot of Lincoln Center. And I was chatting with him after a show, and he was describing the process of describing the process of transcribing one of one of the earlier greats that he loved. I think it was a Sonny Rollins solo that he, like, nailed, literally slowed, used any tune to kind of slow it down without, you know, without affecting the pitch and, like, got it down to an absolute T, a very complex line that he just loved one of his that moved him. And then he kind of described when he was, like, when he finally nailed it down to like, millisecond precision and just tonal, the tonal characteristics of it, it kind of lost all its meaning, like, once he was able to perfectly produce it, right? He was asking his wife, and he was like, why did this, you know? And she's just like, it's sort of, yeah. It's like it lost. It lost that the human element of what was driving that in that moment became it lost its meaning, which is kind of interesting, because it was such a perfect reproduction of it, yeah. But I could see kind of a similar uncanny valley with a lot of this, you know, a lot of this stuff.
PP: Yeah, one of the stats the guy talked about was that, you know, there's, there's hardcore music fans and people who travel to see shows and, you know, go see live things, buy all the albums, box sets or whatever. And then there's the majority of people, which is like, just put a playlist of songs that I like, you know, and that's probably, like he said, 80% I don't know if that's true, but it feels true. And so if AI generated music that sounds like songs you like, is becomes very possible. It feels like that's going to be some kind of portion of music at some point. And you know, it's kind of like, how, how should we feel about that? And to your point, even if it's technically proficient, will people still have that resonance? And because, in other words, could, could AI music. It's less about like, I'm not arguing the 80% statistic. It's more like, what? How much of actual AI music will that be? If you know what I mean, because if it isn't, if done by a human, you don't have those imperfections, that would you even like it, right? You know?
AE: Yeah, it's like watching a robot do gymnastics or something, you know, yeah, exactly, crush a human. But who knows if you know, is that really going to be something that people want to watch at the Olympics, or is it about the humanity of it? You know? Yeah, totally, yeah.
Cedric: I think, you know, I kind of said that on stage. Unfortunately, you can't put it back in the box, because when you do it for the masses, you I believe, like in the generations that grow up on it, they'll, they'll be missing that feeling anyway. They won't, because the way that they'll have to digest the music that's coming their way. So anybody that can figure out, you know, you know, the way to create music, or create a feeling of a song that they love, and and then I also have the algorithm to push it to you first, like you like the example of this is the worst song, but I sold it to you so many times. I now don't hear the imperfection like I just don't hear it. And so to your point, it's going to be a hard thing to keep you know, for people to understand what they're missing. And. Like, what are you talking about? Like you're missing it.
AE: If you think about, like the Motown guys, they would Yes, the Wrecking Crew of musicians that were just they would be in the pocket, they could nail a session, and they would all do it live, yes. And now they would probably those guys from the 70s and 60s rather, would probably look at music production today and be like, this is, this is, this is Pete, you know, like you're not in the room actually creating this music. This is asynchronous.
Cedric: I do a joke about it actually kind of killed the, you know, in a lot of ways, R and B, I mean, in those great bands, you know, almost in all music genres, in a lot of ways, but especially for black music in in the sense that the Earth Wind fires, Ohio players, bands, you won't see them anymore. Everybody now is on computer, and the great music, like I say, everybody's by their self. Drake's all alone. Rick Ross by itself. Travis Scott's by itself. Kanye is by himself. Kendricks, everybody, nobody's in a group anymore. Yeah, no, it used to be a group of people that attempt to make music. All of us come over, be in my band. Come to my house. You play the drums. I do this now. I can do it right here. I don't need you. And I literally put it out. And I'm the man, and that's me and so, and that's the way it is. And so, you know, you kind of miss the, you know, that spirit of what it takes to, you know, put out something that you know for, for the for the purposes of being, you know, you two, or you know, the Foo Fighters, people that you love, you know, Red Hot Chili Peppers, like they, you think about how they hit you when the music comes on, you like, but that's because everybody's different in the group, and they're bringing you something different that's showing up in one piece of product. You know, the Beatles, of course, all these great, great performances.
PP: yeah, I kind of worry about, like, what the guy said that it's sort of when not to use AI and when not to sort of do that, because I feel like you're right, that tapping into sort of that isolation trend, I feel like AI is probably going to exacerbate that. And there I feel like, not just in music, but in a lot of industries, I think about this with journalism and sort of, like making things easier for people early in their careers. And like, well, that's the time when you really should be doing the reps work in terms of the tedious work.
AE: Are you worried? Is it, particularly with journalism and stuff? Is it? Do you Do you feel that they'll just be lacking a muscle in a way?
PP: Oh yeah, I already see that, you know, like I grew up, not grew up, but like, the early in my career, I was, like, in magazines, and they're perfectionists at magazines. Yeah, little thing, because you have a month or a week to prep the thing, and, like, little things like line space, I had a tyrant editor who just completely, like, made me a perfectionist. And I've seen, even without AI, a trend with the internet and online and social media, people have just gotten sloppier, right? You know, there's, there's still need for, like, copy and stuff now AI can kind of clean it up, right? So it's like, it's kind of even worse that we're getting first drafts now and they're not really kind of inclined to do the work in the same way. I mean, I guess it's kind of like different kinds of work. So I do worry about that, yeah, for sure. But I feel like with music too, it's like, like, I think in almost any progression, like, when you're just starting out, you need, you need to do reps, you need to do build the muscle memory, right?
Cedric: And, yeah, you have to do. And I think that, you know, you see that it's also generational, too. I mean, while you have great, you know, kind of advanced thinkers, you also, to your point, you start to build a generation of people who don't want to do they don't want to do the reps, and so then they it just becomes really lazy. Was what the professor was, you know, kind of alluding to, and is that kind of, you know, letting the computer do what I know, that what I don't have to do, but I can take credit for it is a wild space for a human being to be in.
PP: One Last question. You said, friend or foe, I'm looking asking you to forward, looking in the future, utopia or dystopia with AI. I didn't think about it all, like the job loss potentially, and the deadening of art, maybe, or versus, like, unleashed creativity, new jobs and investments created, etc,
Cedric: It could be I’m an optimist by nature. I mean, it's just me. I just, I feel like, you know that the spirit of the human, the the the dynamic, you know, nature of people that I met as individuals will always just, will find its way to the top. It'll just be the. Best, and so I'm more euphoric than what I think about it.
AE: Yeah, perfect. I couldn't say it better than that. I agree. I'm an optimist. I think there's definitely risk. And someone I play the saxophone. I love acoustic music. I mean, there's, that's, there's something, there's something very distasteful about it, kind of like as an immediate, guttural response. But I think ultimately it'll be great for humanity, and it might even, you know, cause future generations to yearn for even more for that which is human. I mean, you think about how live shows now, or, yeah, you know, bigger and bigger thing about each of YouTube, you think that 4k television would make live irrelevant. So what is it that people yearn for in that context. And, you know, hopefully it'll make artists have to, I mean, ultimately, artists will have to do better and engage with it.
And it'll be, you know, if you're just kind of putting out undifferentiated material that doesn't really, you know, then, yeah. I mean, there's a risk there for media. There's a risk to mediocre art, and a risk to, I'm sure, media for journalism and, you know, but I think there's, there'll always be demand for the human element. And so I'm an optimist, because we're humans, after all. So, you know, I think we there's some craving for that which is human, and so I think it'll empower us. And, you know, maybe there's a kid who makes a great sci fi film, and that's a little weird because he didn't have, you know, 3000 people in post production, or whatever, you know, yeah? But you know it's, yeah, it's great
PP: Thanks so much, you guys.
From Drake to Data, with Cedric the Entertainer and Qloo CEO Alex Elias