We’re launching a podcast! Gareth and I will be talking through all things music and tech in seasonal form. The first season is on AI and we’ve got some great people lined up to talk to. Kicking off with Michael Pelczynski. We’re punching through the noise. People write and say that music is at an inflection point. Again. That’s not what you’ll hear here. Instead, we discuss how, as a music ecosystem, we respond to them. What struck me about the conversation is that it’s not really about technology, but about trust. Jump in and see whether we’re asking the right questions in the first place.
The consumption trap
Music has always been a consumption-based industry and we’ve built an entire ecosystem around scale. Currently, that means more streams, more users, more reach. The margins are thin. The question of “who pays?” inevitably lands on either the artist or the fan, and both can only be squeezed so much. Sound familiar? It’s the same structural issue I’ve written about before when discussing why distro is changing and why we need new music economies.
The streaming model gave us unprecedented access, but it also locked us into a framework where volume has become the only measure of value. Now, with GenAI entering the picture, we have two paths: we can try to force this new technology into the existing copyright complex, or we can rethink what monetization looks like entirely.
As Mike points out in our conversation, the old playbooks simply won’t work this time. The streaming royalty model doesn’t fit neatly onto AI-generated content. If we can’t rely on the same licensing structures that we’ve built before, the infrastructure needs to change.
Generative music isn’t just a threat
We often see people have a tendency to view GenAI music purely through the lens of displacement: AI will flood platforms with slop, dilute discovery, and undercut human artists. Of course, this is happening. We keep following along with Deezer to understand how much AI-generated music gets uploaded to streaming services every day. This is fraud and bad actors are real problems that require system-wide solutions.
But that’s not all there is. GenAI music also opens up entirely new business models. Personalized soundtracks for games. Adaptive music for retail environments. Custom compositions for creators who need background music. These are new markets that didn’t exist before because the economics never made sense. This is where Mike comes in. He asks the question whether we can build these markets in ways that compensate creators fairly and maintain the cultural value of music. For him, that requires creativity and experimentation, and definitely not just defensive postures.
Trust is the foundation
One of the most striking moments in the conversation was when we discussed trust. Trust in the people building all the infrastructure and the systems they create. This is where the music industry’s history works against it.
We’ve been burned before. The promise of streaming was fair compensation and transparency. The promise of social media was direct artist-fan relationships. The promise of blockchain was ownership and control. Each time, the structural realities, venture capital demands, platform monopolies, misaligned incentives, undermined those promises.
So when tech companies arrive with AI models trained on music catalogues without licensing that music, without clear compensation frameworks, and when they promise “revolutionary” tools without addressing input/output dilemmas, our skepticism is earned. We just need to understand the power dynamics here.
Skepticism can lead to distrust and that can lead to paralysis. If we refuse to engage at all we cede the entire conversation to the companies building the tools. We need to be in the room - and Mike is - shaping how these systems work, demanding better terms, and experimenting with alternative models.
The input/output dilemma
The licensing question around AI is genuinely complex, nobody is denying that. Do you license the training data? Do you license the output? Both? Neither? The legal frameworks are still being established, and precedent is being set in real-time through various court cases and settlements.
What Mike makes clear is that we need much more granular approaches to rights and permissions. A blanket “opt-in” or “opt-out” doesn’t capture the nuance of how different creators might want to participate (or not) in GenAI systems. Some artists might be comfortable with their work being used for training but not for direct generation. Others might want to participate in specific contexts but not others. The infrastructure for managing these preferences barely exists right now.
This is where collaboration between technologists, lawyers, policy makers, and musicians becomes essential. We can’t solve this purely through legislation or purely through market mechanisms. We need to build new systems together and these are both technical and cultural. The key building element centres on trust and that means respecting creator agency while enabling innovation.
Creativity must lead
In our conversation, we kept coming back to one theme: creativity needs to be at the centre of all of this. Where others focus on efficiency, scale, or automation, Mike brought the focus on creativity. He discussed vocal models and how they are helping artists establish a source of truth. Of course, they’re also another business opportunity.
All these AI tools, generative and other, are getting cheaper and more accessible, but that doesn’t make creativity itself cheaper. Knowing what you want, having taste, developing a point of view are still scarce resources. As Sari Azout put it: “the real challenge is to rethink what we optimize for” when we create AI systems.
Does this sound optimistic?
I keep coming back to a question I’ve asked before: does this actually sound optimistic? It’s about systemic failures, structural problems, and the kind of technology that could fundamentally reshape our already fragile ecosystem.
But yes, I am optimistic because the conversation is shifting. More people are asking the right questions. More artists are experimenting with these tools creatively. More developers are thinking about ethics and compensation. More funding models, such as creative R&D, prefunding, patronage, return on culture, are emerging to support different approaches.
Like Mike said, the old playbooks won’t work this time. Good. Let’s write new ones.
Listen to the full conversation in the podcast episode above, where we explore what it will take to build a music economy that works for everyone.
Have thoughts on GenAI, music monetization, or building trust in new technologies? Join the conversation in the comments below
