Artists and AI: Offense, Defense, and the Future of Music

Summary

Trapital Podcast Summary

Episode: Artists and AI: Offense, Defense, and the Future of Music
Host: Dan Runcie
Guests:

• Kakul Subastova (CEO, Splice)
• Michael Nash (EVP & Chief Digital Officer, Universal Music Group)
  Date: April 24, 2026

---

Overview

This live episode, recorded at Humanx (an AI conference in San Francisco), explores the evolving relationship between artists and artificial intelligence within the music industry. Dan Runcie convenes Kakul Subastova and Michael Nash to discuss:

• The proliferation of AI-generated music vs. human artistry
• Balancing innovation and protecting creator rights
• What the future may hold for both creative and fan experiences as AI further integrates with music

The conversation challenges simplistic narratives of “artists versus AI,” instead framing AI as a creative tool—and dives into how industry leaders are structuring partnerships and new tools to ensure artists stay in control.

---

Key Discussion Points & Insights

1. The Explosion of AI-Created Music

Michael Nash’s Perspective

• The idea that AI is “eliminating music content supply constraints” and “jailbreaking the economics” is founded on a faulty premise. There was already an overabundance of music before AI:
  • "100,000 tracks being uploaded to services every day... If you were to never sleep... you’d have to live 1,700 years in order to experience all music previously." (Michael Nash, [02:44])
  • With AI, “you'd have to live 17,000 more years for every year that AI models are cranking out new music.” (Michael Nash, [03:06])
• The key insight: More content doesn’t mean more value—connection and meaning drive music’s impact.

Kakul Subastova’s Perspective

• The perceived conflict of “artist vs. AI” is flawed.
  • "No one's going to listen to 17,000 years worth of content. What people are going to listen to is the piece of music that connects with them." (Kakul Subastova, [05:14])
• It's always been about tools: From cave paintings to AI, technology is an evolving artist’s toolkit.

---

2. Artist Control and Responsible AI Development

Splice & Universal Music Group Partnership

• Three primary focuses ([06:42]):
  1. Ethical Training: Working out the right way to train on UMG’s catalog—“how we do training the right way… not rocket science.” (Kakul Subastova)
  2. Artist Input: Involving artists directly in building creative AI tools—“build the next generation of creative tools with the creator next to you.”
  3. Novel Fan Experiences: Inventing new ways for fans to engage beyond passive listening.
• Respecting copyright and fair compensation is central: "Tools for artists created by artists, respecting copyright and compensating artists." (Michael Nash, [07:57])

Industry-Wide Shifts

• Early AI in music felt like the “Wild West;” now, more companies are seeking to “do the right thing” around artist rights ([09:09]).

---

3. Balancing Innovation (Offense) and Artist Protection (Defense)

• Universal’s stance starts with “defense into offense”—protecting artist rights is foundational to any innovation.
  • "If you don't claim a seat at the dinner table, you're liable to wind up on the menu." (Michael Nash, [10:41])
• UMG’s approach involves:
  • Collaborating with a portfolio of tech firms, from startups to giants (YouTube, Nvidia)
  • Investing in new AI-patented solutions (e.g., AI-powered sound therapy on Apple Music)
  • Pioneering industry transitions: “From blocking UGC to enabling it,” “from suppressing social expression to enabling it with Meta” ([11:42]).

On Artists’ Desired Control

• Artists’ boundaries and desires vary—some want to create every sound, others are comfortable assembling samples ([12:51]):
  • “For every artist, it's a different line... The sense of something of me is part of the thing that comes out is very, very important.” (Kakul Subastova)

---

4. What Makes an AI Tool “Better” in Music?

Is ‘Better’ About Output Quality?

• Nash challenges the idea that technical improvements mean better art:
  • "Would Bob Dylan's music be better if his vocals were perfect?" (Michael Nash, [15:01])
  • “Art is the triumph of content over form.”
• The unique, authentic narrative of the artist remains central. AI may enhance production, but creativity’s value comes from the “inside out nature of music.”

Subastova’s Vision

• Improvement is about enabling authentic expression:
  • "Ultimately it's the creator tool and how that helps the user tell their story." (Kakul Subastova, [16:35])
  • The AI music market hasn't yet delivered tools that truly allow musicians to express themselves (“super early innings”).

---

5. Looking Forward: AI in Music, Five Years Out

• Michael Nash: Predicts a “paradigmatic transition”—AI will enable “intelligent ecosystems,” ushering navigable, hyper-personalized fan experiences, amplifying creativity, and expanding the music economy ([17:21]).
• Kakul Subastova: Foresees a future where AI in music is simply inferred, not discussed:
  • “We won’t be talking about whether music was made with AI... Just like we don’t talk about is this electricity that’s powering these lights.” ([18:07])

---

Notable Quotes & Memorable Moments

• Michael Nash: "If you don't claim a seat at the dinner table, you're liable to wind up on the menu." ([10:41])
• Kakul Subastova: "There are only two people that matter—the person putting their heart and soul… and the person who is listening… That’s a conversation." ([05:38])
• Michael Nash: "A nuclear explosion of content through AI doesn’t have a market. There’s no audience for that." ([03:34])
• Kakul Subastova: "We do not have the right creator tools yet in market that truly allow musicians to tell their stories. Well, at least in the AI space, we don't have them." ([16:50])
• Michael Nash: "Art is triumph of content over form." ([15:15])
• Kakul Subastova: "We won’t be talking about whether this music was made with AI… we just take it for granted." ([18:07])

---

Timestamps for Key Segments

• [01:31] – Debunking “AI Versus Artist” narratives: Content overload doesn't equal listener value
• [04:28] – Technology as the artist’s eternal toolkit; emotional connection remains supreme
• [06:23]-[07:57] – Building artist-centric AI tools: Partnership structure & philosophy
• [09:38]-[10:08] – Balancing innovation with artist rights: “Defense into offense” in action
• [12:51] – Artists’ need for control and authenticity; personalization vs. automation
• [14:32]-[16:50] – What is a “better” AI music tool? Human narrative vs. technical progress
• [17:21]-[18:07] – Five-year outlook: Invisible AI, personalized music ecosystems

---

Conclusion

The conversation cements that AI isn’t a threat to artists but a new, evolving tool—one that must be wielded with ethics, respect for creators, and a focus on authentic creative expression. The future is less about perfecting sound or maximizing content and more about empowering artists and deepening audience connection. As in past technological revolutions, the winners will be those who put creators in the driver’s seat and innovate responsibly.

---

For listeners with an interest in tech, music, or creative rights, this episode decodes not just industry jargon but also its philosophical and practical stakes—and where AI-enhanced music is headed next.

Transcript

[A]

Foreign. I'm Dan Runcy. Welcome to Trapital. Today you're going to hear a conversation that I moderated on stage at Humanx, this big multi day AI conference in San Francisco at the beginning of April. And I hosted a talk about AI and putting artists in control with AI. I was joined by Kakul Subastova, the CEO of Splice, and Michael Nash, the Executive Vice President and Chief Digital Officer of Universal Music Group. We covered a bunch of topics, but three key themes to look out for. First, the abundance of music created with AI. How does it compare to the music that's already been created on music streaming services and uploaded daily? And what does it mean and not mean in terms of music fans and how people actually relate to the music that comes out? Second, we talked about finding that right balance with AI. How do you push innovation forward so you don't get left behind? But how do you also respect the rights of the people that own the music and the people that make the music and have certain desires that they have to have in their control and what they're willing to let a tool or platform do? And third, what does a better AI model actually mean for the artist and for the fans? Really fun conversation and I hope you enjoyed it as much as we did. Here's our talk at Humanx. So, Michael, we'd love to start with you. What is your thought on that framing and where we are right now?

[B]

That whole framing about artists versus AI, let me just mirror it to its conjoined twin, AI against artists. I think that there's a false premise, false narrative that really comes from an investment thesis that needs to be pulled apart here. There was an idea that I think led to investment in AI a few years ago. The thesis is basically AI eliminates music content supply constraints, kills the legacy model and jailbreaks the economics of the music ecosystem. The problem with that thesis is that we already exist in an infinity of music content. There was no constraint on music content creation prior to the rise of AI four years ago. 100,000 tracks being uploaded to services every day. So let me do a little cocktail napkin math to break this down for you. If you were to never sleep and attempt to listen to all music ever created, you'd have to live 1700 years in order to experience all music previously created. And then you'd have to live 17,000 more years for every year that AI models are cranking out new music. That's the volume of output from just one AI model this year. So for those of you that are planning immortality Based on singularity, you'll never catch up, right? Because you could live 1700 years and listen to all the music. But then every year that you've spent listening to all the music, another 17,000 years worth of music has been produced. So forgive the reductio ad absurdum, but the point is, a nuclear explosion, a production of volume of content through AI doesn't have a market. There's no audience for that. It's not addressing any kind of need. But you flip it back in terms of artists and how they're thinking about AI, how artists are thinking about AI is there's an incredible opportunity to significantly amplify creativity. Artists are interested in and are engaging in the development of and the application of AI tools because they see tremendous opportunity for, with respect to their creativity, with their the compositional process itself, new approaches in terms of music production, and the opportunity to use new types of AI instruments to produce new experiences for their fans. I think that the really important framing here revolves around why artists have always been excited about advance in technology. It's the same thing that applies here. So it's really artist times AI, that AI has an opportunity to be a huge force multiplier. So instead of thinking about going from 1700 years worth of content to 17,000 years worth of content and seeing an order of magnitude explosion in irrelevant content volume, we're looking at it as an order of magnitude advancement in creative potential for artists engage with technology.

[C]

I come at this from an artist's point, point of view, and what is it that the creator wants? So just a quick background on Splice. For those of you who don't know, we are one of the largest music creation platforms for musicians today. We provide them with samples, the sounds, the ingredients to make music. We provide them with virtual instruments and new creative tools, like Michael was just saying. And we're building new AI first technology to give users and our creators even more power, even more creative control. And we have a partnership with umg. And so the idea that there is an artist versus AI thing is actually deeply flawed. I actually believe that technology has been part of the artist toolkit from the beginning. 40,000 years ago, when people were making the cave paintings, they were using the technology available at the time, right? Smoke, fire, motion to get a certain narrative across. And today AI is just the newest tool that's in their toolkit to help them tell their stories better. Because as Michael saying, no one's going to listen to 17,000 years worth of content. What people are going to listen to is the piece of music that connects with them. And there are only two people that matter. The person who is putting their heart and soul, regardless of what tool they're using to create a piece of music, and the person who is listening to that piece of music and is having an emotional reaction. That's a conversation. And that conversation is just as true today, regardless of what your tool chain is, than it was four years ago or 40 years ago or 40,000 years ago.

[A]

And Kakkul, you mentioned that Splice and Universal Music Group now have a partnership which is recently formed as of a couple months ago. So based on what both of you shared about putting artists in control, what does that look like for you all? With this partnership and in practice, there

[C]

are three primary things that we're working on together. The first is working with Universal to really work out what's the right way to train on the incredible catalog that Universal has. So we're having some conversations around exploring how we do training the right way as opposed to the wrong way. And it's not rocket science. And it's been a really great partnership and a really great conversation around that. The second thing that we're working on is how do we bring some of the incredible artists that are part of the Universal Music Group's portfolio to the seat at the table with us to help us create novel experiences? At Splice, we really believe that you have to build the next generation of creative tools with the creator next to you. And so how do we do that, brokering that conversation then? The third thing is can we invent new ways for people to engage with music? There's one thing that's passive listening. They're more forward looking things that users want to do with the music. How can we brainstorm together to invent some of these new ways? All three of those conversations are active and moving forward.

[B]

I would just add that what's really attractive to us about the process that you just outlined is this is really tools for artists created by artists, respecting copyright and compensating artists. So it's an all inclusive formulation and we see a great opportunity with artists in the driver's seat to really understand requirements. Gathering, workflow, calibration, what I would call feature set parameterization, like how you identify the artist or a certain style in relationship to a desired output. The process that you've designed that we're going to work collaboratively to advance is well positioned to achieve the right outcome and to create a lot of engagement by the artist community. And the other thing I would say just about what attracted us to the partnership in terms of values and vision. We worked to establish principles of music creation with AI, a kind of artist centric set of guidelines several years ago. And Splice was one of the first signatories. Splice is run as a very artist, first artist centric company. And so the opportunity for this partnership to focus on a collaboration to develop something that I think we both identified as a key objective with Splice made sense from the beginning.

[C]

I'll just say that it's not just Splice. I think the next generation of AI companies are fundamentally trying to do the right thing where it comes to respecting the artist rights and copyright and legal frameworks. It's not rocket science and you really are seeing that. Early days were different. It was a little bit Wild west, but we are one of many companies that are trying to do the right thing.

[A]

Let's talk a bit more about respecting artist rights. Since both of you mentioned that Universal Music Group does represent some of the most talented musicians across the world. And Michael, I'm sure you've probably heard a wide range of stances from artists that Universal Music Group represents on how they feel about AI. So how do you balance the offense of trying to keep this company innovative with the defense of respecting the rights?

[B]

Well, we're obviously extremely oriented on the rights and interests of our artists. And so the starting point for us is kind of defense into offense. And we have an obligation and a responsibility to defend the rights of our artists. But that's also the key component of organizing a marketplace for ethical entrepreneurs to develop new tools and new experiences working with the artist community. So you certainly can say, as we're fond of saying, if you don't claim a seat at the dinner table, you're liable to wind up on the menu. So it was important to initiate a conversation from a position of strength, but the philosophy was really about focusing on innovation. So that defense into offense approach is how we were able to establish a framework for a partnership like the one we have with Splice. We established partnerships with a number of different companies. Bandlab, Clay, Udio, Stability, AI to significantly scaled firms like YouTube and Nvidia. With respect to leaning into innovation too, I think to demonstrate to artists that we're focused on devoting our resources to also create solutions, we've developed a number of AI patent applications. One good example would be in terms of health and wellness, we've developed a process that's enabled by AI which supports our exclusive participation in the sound therapy category on Apple Music. That relates to a broader philosophical stance where business model innovation is really required in order to create alignment between the tech community and the creative community. We've looked at business model moves, whether it's at the beginning of this century. Unbundling the album to enable Apple to create a new proposition for consumers around digital or with respect to dealing with file sharing and enabling a Spotify with a streaming model or with UGC on YouTube rather than blocking UGC. Coming up with a model that basically supports fans engagement with artist content, which is going to be significant from the standpoint of giving us precedence to rely upon in AI. You know, we're in the social space. The deal that we did with meta going from suppressing social expression through music to enabling social expression through music. So I think it's really incumbent on us to figure out how to take the strong position around protecting the artist's rights and interests as a foundation for business model innovation to work to organize a marketplace for ethical entrepreneurs.

[A]

And I think similarly, Kakkul, you all are also balancing artists and AI, specifically where AI can help reduce friction or make things more efficient. But are there certain things that artists should keep in control or things that they've told you? Especially those that may be using splice?

[C]

We've spent years working with artists on these tools. It's actually a conversation that's happening between the artists and the tool set. When we launched our tools a couple of years ago, some of the feedback that we got right away was it's actually a little too easy. Can you give us more control? Because I don't feel like enough of me or my vision or my narrative is showing up. You're not giving me room to put that in because you're just creating thing for me. I think that sense of something of me is part of the thing that comes out is very, very important to artists and creators. For every artist it's a different line for someone. It's like I've got to have made every single sound that's in this track. And for others worth of hip hop sample based music production, you want to use bits and pieces from other places, then there's toil. I love how this thing has come together. It's got this beautiful drum section that I love. I love this vocal piece right here. I love the outro and how it gets really soft over here, but make it all sound good together. And when you have that kind of stuff, you're like, you could probably fiddle for hours to get it to do that. But can we make that easier for you? Can we make some of the Toil easier. That's what we really think about a lot. It's a rapidly changing conversation as both the tools get better, but also people get more comfortable using these tools and finding where it is that they want their voice in the conversation.

[A]

We should talk more about tools getting better and what that actually means. Because of course, Anyone that uses ChatGPT or Claude, we know what it means. When a tool gets better, the output improves, it can relate to your history, we've seen that. But in music, is it that the music itself gets better? What does it mean when an AI tool in music gets better?

[B]

So at the risk of giving you an entirely non technical answer to your question and do it in the form of a question, I would say better music. Better in what way would Bob Dylan's music be better if his vocals were perfect version that question for like Neil Young or Leonard Cohen or Johnny Cash or Kurt Cobain or Claire Courtney Barnett or Lana Del Rey or Billie Eilish or basically any punk rock record. I think that Robert Creeley, great poet philosopher, got it right, to paraphrase him, when he basically said that art is triumph of content over form. When you think about music in particular, it comes from a really deep place. I love how Bono expresses it. Bono's expression is we sang before we spoke. And what he means, if you know a book like the Singing Neanderthals, like academic treatise research stands up that supposition that music preceded language. It's the language of emotion, that inside out nature of music. It kind of works against the idea that technical perfection is the goal of artistic expression. So I think it's technically correct that the music models are getting better and better. But to say does that mean that you're going to see the creation of better music from a mechanistic point of view, that's the only way you would arrive at a conclusion that yes, you're going to see better music. It all relies on the artist and the creative process and artistic intent.

[C]

I do think the technology is getting better. And that's what I mean. I've been in the technology industry for a while now and that's what makes it so fun, is how quickly it all gets better. And no matter how good it gets, the goodness of, of the music that is created will come from the authenticity of the creator's narrative coming through. And so models and agency, platforms and all of those things are fantastic. But ultimately it's the creator tool and how that helps the user tell their story. And I think we're still in the super early Innings of that. We do not have the right creator tools yet in market that truly allow musicians to tell their stories. Well, at least in the AI space, we don't have them.

[A]

Well, before we close things out, we're still building the plane, we're still early. But five years from now, one sentence from you both, what will tell you that we got this right with regards to AI and music?

[B]

The one single sentence constraints kind of difficult, but the way that I think about it is if you consider what the electronic revolution meant a century ago and you consider what the digital transition meant at the beginning of the century, 25 years ago, I think that AI is going to represent a paradigmatic change, a paradigmatic transition in the arts and in music. And five years from now, I think you're going to see the beginning of a transition where artists are going to populate intelligent ecosystems with incredible navigable, hyper personalized experiences for consumers. You're going to see a significant amplification of creativity in music and you're going to see significant growth in the music economy as a result. That was like five sentences.

[C]

Sorry, I'll keep it to one. I really think that we'll be talking about the tool set and how the technology is getting better and those of us in the craft, we'll be talking about that, but we won't really be talking about whether this music was made with AI or not. It just will be how it's done. Just like we don't talk about is this electricity that's powering these lights. We just take it for granted. And I think that's what's going to happen with AI and music as well. As long as we get the right things right, which I know we will.

[A]

Kakuo, Michael, thank you both and thank you all. Appreciate it. And that is a wrap for our conversation at Humanx about putting artists in control with AI. Thanks again to Michael and Kakul for joining us. Thank you again to our audio and video producers, G and Eric, for everything that you do to help make trapital possible. Thank you to Humanx for hosting us, but most importantly, thank you for listening. If there's one person you know that would really enjoy this episode that we just put out or any of the conversations that we have on trapital, send them a link to the show. Word of mouth is still the best way to grow. And if you have a few moments, if you can leave a review, leave a comment that helps the algorithm do its thing and make sure that trapital reaches the right people. Thanks again. Talk to you next time.