“You’re Not Being Creative, You’re Being Algorithmic”—will.i.am Talks Shop About AI

DMN Rules for AI conference will.i.am
  • Save

DMN Rules for AI conference will.i.am
  • Save
Photo Credit: Evatt Carrodus

Digital Music News explored the evolution of AI in a recent panel highlighting how the technology has shaped the industry in 2023 with its ‘Rules for AI’ conference. The viral ghost-written song “Heart On My Sleeve” started the conversation on how to utilize AI ethically and responsibly, with will.i.am quickly reframing things conceptually.

Utilizing artificial intelligence in the creative aspect of the industry should come with guiding principles. DMN’s Chief Revenue Officer (CRO) Noah Itman discussed those principles with our panelists, highlighting how AI can be used ethically and responsibly by artists and creatives within the industry.

Perhaps most notable is the idea that creation shouldn’t be limited to the narrow scope the industry has carved out for itself in the last 100 years of existence.”If I had AI [back when I got started], I wouldn’t use it in the narrow scope of what’s happening today. I would create a whole new music industry,” will.i.am shared with the panel. “The concept of structure was the first algorithm to fit the industry’s lacquer limitations.”

“Even now, we’ve condensed it down to algorithmic limitations in the music industry, this is how to go viral on TikTok! You’re not being creative with that, you’re being algorithmic,” will.i.am continued. He highlighted how he uses AI in his messaging tool for creatives to outline and strategize how content should be released—rather than creating wholesale with AI at the helm.

Meanwhile, TuneCore’s Andreea Gleeson highlighted some of the guiding principles that should define how AI is used creatively.

“Often when there’s new technology that emerges, some people look at it as an opportunity while others say, ‘what about bad actors?’ These two perspectives are both valid and true. I think we’re definitely living through that today and what you can do when exploring new technology is to think ethically about the principles that should guide its use,” Gleeson explained. “As we started talking with Daouda at CreateSafe, it was our first real life example of how we’re going to help artists engage with AI in a responsible way.”

Gleeson outlined four principles for engaging ethically with AI including Consent, Control, Compensation, and Transparency. Obtaining consent from the original artist, giving the original artist creative control, compensating for the use of their voice, and transparency with DSPs when distributing AI-created content by labeling it as such.

“Grimes was very smart and thinking a few steps ahead in training her model on music she 100% owns,” Gleeson explained. “Artists can then use that model to transform her voice and distribute the music through TuneCore. Because Grimes has said we can do this, we have her consent. Distribution credits Grimes.AI for these tracks, giving her a 50% split of any revenue; that’s compensation. The tracks are also routed through Grimes’ creative team, who can reject or accept any track—that’s control. Finally, the transparency is listing Grimes.AI as a creator, not Grimes herself.”

The panel also highlighted how AI can be used during the collaboration process—not specifically for creation. will.i.am shared that creating is “like therapy for me.” He compared using AI in a creative aspect as like asking a yoga AI to do stretches. No matter how creative the AI is at doing those stretches—you receive no benefit unless you do them yourself.

“As far as focusing my ideas and strategizing with collaborators, I use AI for that,” Will shared. “The thinking through how to put [a track] out in the world, what does it mean to the world? It’s more solution oriented for helping with marketing and I like AI for that. I know it has limitations, that it hallucinates. But I love the banter of what it does for my brain to look at a problem uniquely.”

Will shared that during the COVID-19 pandemic, he envisioned a creative messaging platform without limitations. A place where creatives can ask AI to summarize chats, view uploaded files, and collaborate together seamlessly without being spread across several tools. FYI.me is the brainchild of that desire, giving creatives a place to share ideas with feedback from AI for quick summaries and ideas—but without any stake in the actual creation process.

Another nascent use for AI within the music industry is voice cloning. Voice cloning can be done using a relatively small dataset to essentially match someone else’s voice. For example, taking a voice and applying a filter to make it sound like will.i.am. Jordan Young says the ability of AI to clone a voice is just one example of how technology is evolving.

“If you think of the history of music and the way it has evolved genre-wise, music has always evolved alongside the technology of the day. From the electric guitar, to synthesizers, drum machines, samplers, and even auto-tune—each of these innovations created a new sound we’ve never heard before. They shaped how we hear music today,” Young continued.

“That’s how I think of these generative AI tools. These are new tools for new artists to create new music that we’ve never heard before. There is a line to draw though—at what point do you build a tool that completely removes the human element from creativity. We’re really excited about these creative tools, but we as creators don’t want to build a tool that removes the human from the process.”

What about the uses of AI outside of the creative and collaborative processes? Philippe from MatchTune shared how his company is identifying covers on YouTube and other platforms to detect unauthorized cover songs.

“Generative AI is more like a revolution,” Philippe shared. “Every revolution comes with the good and the bad. Generative AI is going to become even more clever in the coming years. We already know that the traditional way of detecting songs and copyrighted content based on fingerprinting is outdated. It’s what we can see today, but it can be tweaked a bit and suddenly there are thousands of covers on YouTube undetected.”

“For example, we scanned nine different networks for a single artist, single track. For this example, let’s say a Bruno Mars track. We found 210,000 videos featuring Bruno Mars that are supposed to be monetized for the artist, but they are not because they are undetected by current fingerprinting technology. Some of them are even monetized to someone else. When you multiply by artists and tracks, this amount of unauthorized covers is massive. MatchTune has a great background in detection using AI tools and as they get more powerful, it complements our mission to find uncollected money for artists.”

With generative AI making it possible for ghost-written songs like “Heart On My Sleeve” to go viral, Noah asked Nick Minicucci whether there should be higher restrictions placed on companies and DSPs themselves, rather than individual creators. The short answer is yes.

“It’s one of the most important things we need to figure out in the music industry right now,” Nick said. “Baseline regulations have to be implemented, but sometimes that gets roped into stopping Big Tech. I don’t know if that’s the best way to think about it. We have to get the foundational elements taken care of now. I agree with TuneCore’s principles of AI, but we have to be ready to evolve because AI will evolve quickly.”

Andreea Gleeson also shared some stats taken from a survey TuneCore did with artists who utilize the service to distribute their works. From 1,600 respondents across hip-hop, EDM, Rock, and Pop—57% have been in the industry for over 10 years. TuneCore asked those respondents if they are familiar with AI or using it already. 50% said they currently use it, or would be interested in using the tech in the future. The most cited uses now include generating creative assets for marketing on social media.

However, 50% of respondents said they would opt-in to machine learning if it was done responsibly. What does that mean? Counteracting plagiarism and receiving proper credit for one. 33% of respondents said they would be interested in utilizing generative AI in their creative process. So the interest in AI is there—as long as its use can be done ethically.

Finally at the close of the panel, Noah asked Jordan Young of Hooky if the service would enable someone to create new works from the voice of someone who has passed away.

“Some artists it makes sense for, and others it does not,” Young responded. “I would never envision Prince wanting his voice to be used in this way. But there are lots of artists that may, if their estates are okay with it. Is it really a huge leap or step too far if Sony Music and the Estate of Whitney Houston want to make a tribute album to an artist we love, with producers whom she would have wanted to work with in her life? To me, that seems like a reasonable approach.”

“There’s a grey area that you have to approach in a holistic way, but I definitely think there’s an opportunity. AI can be used to fill in the gaps, but it can also be used to access entirely new fanbases. For example, one of the things we’re working on is translations of songs. You could have Will rapping or singing in perfect Korean and delivering that song to new audiences around the world. You could do that in any language and have it sound perfectly like Will.”