A Key Change for AI, Music and Advertising?

With all eyes on generative AI, much of the focus has been on its impact on visual creativity. But, of course, artificial intelligence of stripes will have as much of an impact on music as it will on still and moving imagery. There have been some notable experiments in the space, such as the short-lived fake Drake and The Weeknd track that was quickly yoinked from social media and streaming platforms. In a corner of the creative industries, where copyright lawsuits and accusations of plagiarism have been common occurrences before the onset of AI, and where artists have struggled with the impact of streaming platforms on revenue streams, artificial intelligence will likely exacerbate and magnify existing issues and tensions. But, there’s no doubt that there’s also a lot of fun to be had too. 

This week, the Financial Time reported that Google and Universal Music are in early stage talks to licence artists’ music and voices to feed generative AI engines. On the one hand, it offers up a way forward that addresses at least some of the legal, financial and ethical issues around AI-generated music, and opens up new creative frontiers… but equally, it might unleash a tidal wave of sameness.

While the report says that the current focus is on enabling fans the play with and pay tribute to their favourite artists, where fans go, brands are sure to follow. So what does it all mean for brands and advertisers? What opportunities and pitfalls lie ahead? We caught up with the ad industry’s music experts around the world to get their insights into what it all means for commercial music. 

Alberto Farinas, managing partner, Animal Music

I think the general consensus on AI is one of fear. The lack of thorough understanding of AI makes people think it's the end of the artist, but it really is just another tool in the chest. I think AI is almost synonymous with the electric guitar - it's going to bring about a whole new world of music.

As with all new innovations, there’s a moment of excitement that takes over and you kind of fall in love with the opportunities you imagine will come from it. It’s almost like watching magic for the first time and thinking, ‘how is this even possible?’. However, after hours of tinkering and exploring the many and various tools available today, you quickly pass the honeymoon phase and start to realise the reality of it all. You begin to fully understand its potential uses and limitations, but also the tremendous potential it has to become something even greater with the right resources.

When you finally understand how a trained model works, and how machine learning lets us execute selected tasks intelligently and quickly, you begin to see the real potential for expanding our toolset, especially in music and audio. One of the biggest challenges I encountered when exploring the different AI music generators available was the lack of quality and creative taste. Most of the music generated was of poor quality and or lacking any emotional cues. I quickly realised the problem wasn’t the machine’s ability to execute the tasks, it was the quality of the schooling. Most of the models had been trained with countless hours of royalty free audio and music, and so, of course, it’s only able to reproduce what it learned from. It made me wonder, what if it had access to all the music in the world with no limitations - would it then be able to create music that was undeniably human?

I think this is the crossroads of where the theory and math has been proven, now what do we teach it? For obvious reasons, this is where the copyright issues arise. Should these AI models have access to using existing music without a proper licence?

We’ve been using AI as a tool in our creative process for years now. There are many tools that use a form of AI, even if it's just as simple as a music tagging tool. As a music and audio post production company, we need to lean into new tech and innovations constantly to keep with trends and timelines. It's not only exciting, but it’s necessary to keep evolving as a company. The more we can automate and be expeditious on the technical side using AI tools, the more time we’ll have to innovate and push the limits of our creativity. There’s no AI tool for that… yet.

Bruce New, composer agent, Air-Edel

While there are already many videos of AI-produced tracks, such as recordings of ‘Artist A’ performing a song by ‘Artist B’, they are still not real. However, AI tech and those who control it are getting better at producing a result which is less obviously fake. 

Producers of new advertising productions (but this also concerns those in the TV, film and video game industries) are limited by choice. If an instrumental version of a track doesn’t exist, but one is needed, and the original recorded stems are not available to create a good quality version, is it better and more authentic to create a new version using live musicians - possibly original band members - or cheaper and quicker to use an AI programme, and who is actually making the choice? 

In advertising, normal music creation on the basis of a brief, referencing styles and instruments gives a large grey area where care has to be taken to not infringe on copyrighted music. AI-produced music must follow the same legal framework. Use of AI needs to be legislated so that issues of copyright infringement can be filed and completed. The worry is that those inexperienced in musicality are asking the AI to create a track ‘like’ or ‘close to’ their reference, without the experience to musically question the AI result. 

Further impact may be seen in the budgets the brands and agencies put towards their music choices. The cost to licence a pre-existing commercial track or creating a new bespoke soundtrack versus the cost to use AI. 

Hopefully, the public, brand and ad agencies will continue to see and hear the benefit of using real composers, musicians and music in their productions. 

Dennis Culp, executive producer, Singing Serpent

When I think of publishers licensing artist's vocal and instrumental sounds, it brings up all sorts of questions when applied for advertisements.

Use of someone's voice is a direct form of their endorsement - and if I were the artist being ‘used’ or ‘sampled’ I would want significant compensation for my endorsement. Without artist consent for each specific placement, this isn't going to work - ask McDonald's how Tom Waits feels about them using a sound-alike of his voice. (Answer: Tom wins.)  

The whole thing seems like a neat trick, but an expensive one. On the same note, AI-generated music is, by definition, derivative of existing work. Who is going to guarantee that a composition is original? I'm anticipating a lot of litigation in the next few years that will clarify how applicable AI composition and sound-alike samples will be to the advertising industry.

Jennifer Fife, senior producer and music supervisor, Sixtyfour Music

This is a new incarnation of the copyright conundrums that have always existed. The reuse or re-interpretation of existing material presents many creative and financial opportunities, but without copyright protection, artists have less motivation to forge a career in the industry. Copyright law isn't perfect, so with AI, we will continue to navigate this fine line so we can reap the benefits while offering protection where needed. 

How we do it is hard to say, because I don't think we can say yet exactly where AI will prove to be most useful (or disruptive) in music. But, I'm fully on board with exploring the potential of the technology while we figure it out. The industry will adapt, and some great work and fan engagement will come from it.

However, I think these protections are even more important in advertising. We must be especially sensitive to the way brands can potentially infringe on copyright, whether it’s intentional or not. If there’s consent, proper guardrails and adequate compensation, it seems fine, but I wonder if one could create a tool that alters music just enough to avoid infringement while still leveraging copyrighted source material. Much harder to argue for the creative potential in that instance.

Joel Beckerman, founder and CEO, Made Music Studio

Generative AI provides previously unimaginable creative opportunities to connect and create culture, especially in partnership with music artists. Grimes’ ‘Elf Tech’ and the artist MIDNATT’s simultaneous six-language song released using voice AI tools inspire us! But, generative AI tools are leaps ahead. Instantaneous content creation to organically expand our campaigns? Incredibly exciting… if we adhere to some basic principles and values. Here are mine:

The necessity of the human touch: While AI can iterate and scale brilliant facsimiles and ‘fakes’, it will always be the human touch that will bring the magic and make it matter. What will that look like?

Don’t steal, give credit: The artist IP licence is the starting point, but from what sources was the neural net trained? Are all human contributors being credited and compensated? What about in real-time content? Brands who celebrate human creators will be rewarded by consumers

Protect brands in real time: Generative AI doesn’t recognise authorship, copyright or trademark. It knows nothing of cultural sensitivities, appropriation, or context. Consider that ChatGPT is refined in real-time to address and solve. We must do the same.

Accept what we don’t know: No one fully understands exactly how this tech actually works yet. Let’s bring humility to the process.

Generative AI is nascent and chaotic. There will be spectacular successes and brilliant failures. Let’s take big, well-considered risks to create new kinds of commercial art. Our audiences will love us for it.

Jonathan Watts, head of production, Adelphoi Music

The collaboration between Google and Universal in offering artist voice licensing for AI-generated music presents a promising stride forward. This move could redirect deserved royalties to artists and allow them to opt in or out of having their AI voices and creations featured on platforms like YouTube. However, implementing this seemingly straightforward concept is bound to be more intricate in practice.

The crossroads of AI-generated music and the prospect of artist voice licensing introduces captivating possibilities and intricate challenges to advertising. Integrating familiar voices into AI compositions could amplify brand recognition, evoke nostalgia, and fashion unique campaign soundscapes, ushering in a realm of creative potential. For instance, envision the estate of Bob Marley contemplating the inclusion of his AI voice in a fresh reggaeton arrangement.

While the framework around AI-generated music's copyright may require refining, the broader notion of protecting artistic integrity remains valid. Extending legal safeguards against the misappropriation of AI-generated voices aligns with established norms for music licensing. Granting artists veto power over their AI voices echoes the process of approving licensed music.

Conversely, reliance on AI-generated music might dilute the emotional resonance and authenticity intrinsic to human-crafted compositions. The visceral connection music forms with audiences could erode if AI homogenises its diversity, potentially muting the impact of advertising campaigns. Striking equilibrium between AI's efficiency and the invaluable touch of human creativity becomes pivotal.

Music companies can harness AI to streamline production, expanding their creative possibilities for more refined outputs. Still, cherishing human artistry as a distinct value proposition is paramount. A nuanced strategy could involve AI-generated music complementing, rather than supplanting, human-crafted works. The horizon likely entails a fusion of both, where technology elevates human ingenuity, enriching the auditory landscape while preserving the enchantment of meticulously crafted human music.

Lauren Thackray, client services director, Big Sync

The evolving landscape of music and technology necessitates a balanced perspective that welcomes AI integration while preserving the distinctive value of human-crafted music. Brands are responsive to the ‘sound on’ economy and understand the importance of staying culturally relevant.

In this regard, human-crafted music remains paramount, as affiliations with cultural icons facilitate a deeper connection with audiences. While acknowledging the significance of established artists, there is also a growing focus on nurturing partnerships with emerging talent across diverse niches. This approach not only champions authentic storytelling, but also aligns brands with emerging cultural trends.

Collaborative licensing structures, strategic brand affiliations, and the cultivation of emerging talents all contribute to fostering meaningful collaborations and authentic narratives that serve as the foundation for sustained campaign success.

Pierre Carnet, managing director, MassiveMusic Dubai.

At MassiveMusic, we prioritise addressing these key challenges as we focus on creating distinctive musical value for brands. As bespoke music creators, we’re looking into AI tools to expedite our production process while preserving human involvement for crafting high-quality music.

On the one hand, AI’s ability to replicate voices offers new forms of artistic creativity. Think of composing with custom lyrics sung by unexpected pairings like 50 Cent performing blues, or a collaboration between Lil Pump and Johnny Cash! This extends to brands owning adaptable voices for consistent communication across all of their touchpoints.

Yet, AI-driven music creation poses competition for creators, and raises concerns about IP protection and monetisation. While existing laws in most countries shield melodies, lyrics and chords, safeguarding thousands of AI-generated variations and preventing unauthorised usage presents challenges. It’s clear that regulation has to catch up with a framework to protect music creators.

Unfortunately, history has shown that new technologies are rarely limited by regulation in the creative space. So perhaps, instead of constraining AI music growth, focusing on managing outcomes and leveraging it for enhanced creativity might seem a more pragmatic answer.

As long as the organic and human element of music remains, we can easily imagine a world in which human and AI-created music cohabit peacefully. Let’s make sure we don’t lose our voice, and use it to amplify it instead.

Simon Elms, director/composer, Eclectic Music, Bark Soho and Syncbubble

If this interest in vocal AI was driven by artists – by singers – then I would be interested in it. Similarly, if it made big Gary from the drycleaners sound like Chaka Khan on karaoke night, it could be a fun thing. However, the cynic in me thinks this is another record company ‘suit’ driven ploy to squeeze more money out of their artists, and if that is the case, it needs to be forensically dealt with. Luckily, Bette Midler has done most of the heavy lifting on this. In the late ‘80s, she successfully sued a very established advertising agency and one of the biggest car companies for imitating her voice. The judge at the time ruled that our voices are as individual as our fingerprints, and any attempt to impersonate them is an infringement of our civil liberties. That’s a good thing. 

Having said all of that, I could imagine that advertising could really benefit from it as long as all of the permissions were in order. Mimicry has always been a cornerstone of the advertising world, and it’s often at its best when it recreates famous film scenes or re-records well known classics. It’s such a thing. I could just imagine Haribo replacing the kids’ voices with the ‘Rat Pack’. You can have that one for free, but generally, imitation or impersonation is nowhere near as popular as it was when I was growing up. In the ‘70s/’80s, impersonators were big business. These days, I think people crave originality - character. Things that are tactile. That’s another good thing.

Rani Zarina Vaz, managing Partner & executive creative producer, Supreme Music

Musicians will continue to be at the forefront of the mix of technology and humanity, pushing boundaries and developing new means of expression through creative innovation. Assistive AI has long been integral to music production. Here we are, just four months since ‘Heart On My Sleeve’. From April’s cease and desists for the use of unlicensed materials to today, talks between Universal and Google signal forward momentum. This is a pivotal juncture, when it is imperative to develop a common language around AI, along with ground rules for inputs, safeguards and accountability. Generative AI can be used in ways which are both ethical and protective of artists, but it’s still the wild west out there in terms of process. While the ‘Times’ article specifies ‘tools for fans’, looking from a purely production-centric view and jumping ahead to the time when the sticky use parameters have been sorted, we see these developments as tremendously exciting. Specifically from the standpoint of using copy-written materials, just imagine the myriad creative possibilities. Generative AI, with human guidance, will also give producers, composers and mixers the ability to tackle some of the more practical and ‘invisible’ production challenges which occur during post.

Photo by Possessed Photography on Unsplash