Man vs machine, challenges vs opportunities, regulation vs innovation.
The music industry is on high alert over AI and, as the debates roll on, the pendulum swings this way and that in our effort to find the sweet spot where we can both capitalise on and control the tide of generative AI.
As someone whose daily round is the digital licensing of musical works, I want to make the case for another spectrum for guiding our strategy. At one end is Culture; a C-word that I wish was used much more frequently in our industry discussions. At the other is Content; a C-word that is pretty prevalent. For my money, Culture and Content are two cross-cutting descriptors that can help us signpost the commercial value gap between human-created and AI-generated works.
Culture is the fluid and evolving product of continuous human interaction. It’s the way we discover and express the meaning in our lives and, though it takes many forms, music is its pulse. For millennia, we have shared our ideas and feelings through rhythm and song.
We have made instruments to express and enhance our creativity and have found ways to share and record our creations. Musical works that come from the culture can be magical, story-telling time machines that shape our identities, our communities and the emotional soundtrack to our lives. We love them, value them and they have huge social capital.
Content, in contrast, simply fills time and space. It can be of high value or low value and is usually measured quantitatively. Content can be multiplied, prorated, sliced and diced. Culture, on the other hand, can only ever be truly understood qualitatively.
So, why is this distinction useful to the AI discussion? Because it helps us to orient our strategy in a positive way by focusing on the dense cultural value of human generated works rather than the hollowness of AI-generated works. That’s not to say that AI generated works can never have cultural value.
They can contain some of that unique energy if they are powered by meaningful human intention or if they spark people into creating stories around them. The critical point, though, is that they have to have meaning given to them. They don’t come with it. From a licensing perspective, another way of expressing it is that human-generated music is both Culture and Content whereas, with some honourable exceptions, AI-generated music is only ever Content.
One practical example of how this framework could be useful is the question of tagging. For very good reasons a consensus is developing around the desirability of classifying and tagging AI-generated musical works and recordings. However, a Culture / Content approach could encourage us to look the other way down the telescope and consider whether we should not instead (or as well) classify and tag human-generated music. This would create a kind of human ‘hallmark’, generating a prima facie elevated status for the song or recording as a cultural artefact. Within a collective licensing environment, works of Culture could then be distinguished from works of pure Content and given a value that corresponds to the way that their cultural capital attracts clients and users to the licensee’s service.
In the publishing world, a tagging system already exists through the writer IPI numbers. These designate the claim that a registered work is a human creation capable of copyright status. In the record industry, implementing a hallmark would require more thought and effort because of the widespread use of AI tools by human producers and engineers.
However, if we choose to look for it, we can probably find some kind of demarcation that we can live with. After all, we have lived for many years with fluid and contestable concepts like “originality” and “substantiality” and we have built workable systems around them in order to handle commercial exploitation. Of course there are nuances, especially for composite works. However, taking the trouble to claim an elevated status for human-created works that are part of our Culture puts us in a very different place than expending our energy on downgrading the status of AI works, the generation of which is largely out of our control.
“Content can be multiplied, prorated, sliced and diced. Culture, on the other hand, can only ever be truly understood qualitatively.”
As we know, AI is here to stay, and I think we can assume that AI-generated music will have its place in the commercial realm. It is already consumed and appreciated in certain contexts and, in the streaming environment, it already competes with human-generated works to some degree. However, it does not appeal to the ‘lean in’ listener like human-generated music does because it is not part of the Culture and it does not carry with it the myriad pockets of meaning that connect us to ourselves and to each other. Yes, we may get confused for a moment because AI can generate some likeable vibes.
However whilst AI can fake sounds, it can’t fake Culture and, quite quickly, we will gravitate to the music that reminds us how we felt when we went to our first gig, or where we were when we had our first kiss and we will continue to share the sounds that connect us to our social “tribe”. Let’s face it, a streaming service comprised entirely of AI-generated music would have low appeal. It’s probably quite nice for a moment but it’s an ersatz experience….like an online concert by an avatar.
Yes, the online experience has something to it and the sound quality through your headphones is probably fabulous. However, it can never compare to a live event even if the gig breaks the bank, involves drinking questionable beer and has you catching the last bus home. The fact is that we’re are happy to pay and to go to all that effort because live music is real and it makes you feel something. Translated back to the digital services, it’s the intrinsic human appeal that delivers a premium experience and draws users to its flame. This is why hallmarking music that we deem Culture is worth the effort.
Making the argument for a Culture premium will doubtless generate pushback from the DPS and other music-using businesses who will find it tempting to plump up their offering and dilute the licensing pool with cheaper Content. However, we must be assertive. Changing licensing models takes effort, time and collective argumentation but it can happen. We are already seeing the emergence of artist-centric models in digital licensing. So, why not also a Culture-centric approach?
Not only is a Culture-centric approach a way to leverage value as a matter of simple market logic, it’s an approach that can give us more control over what’s to come. At the moment, CMOs are not registering AI Content… but the collective licensing industry may ultimately decide that finding some accommodation for this might be in our interests.
After all, playing a key role in a single overarching Culture / Content licensing scheme with differentiated rates might be a better outcome than separate schemes running along different tracks where we are only influential in one. If that’s the case, then having a positive hallmarking strategy for human-created music makes a decision to register AI works much less of a risk. Our own hallmark means we can easily identify everything that is not in the Culture bucket without having to rely so heavily on the bona fides or accuracy of tagging by the pure AI Content providers.
As the CEO of an international collective of independent music publishers, I am proudly in the Culture business. Advocating for Culture and securing its value is what we do. Focusing on that positive task rather than on the fear of a tsunami of Content is surely the best strategy.Music Business Worldwide
