Artificial intelligence has become a hot topic for many industries worldwide, especially in music. As it has continued to develop rapidly over time and become more convincing, many questions have been raised about the sustainability of both the tool and the artist.
At the seminar at Indie-Con 2024, five local, interstate, and international guest speakers from music and political backgrounds provided a global perspective on AI’s potential threats and opportunities.
The first guest speaker, Senator Sarah Hanson-Young of the Greens, emphasised the need for the local government to support the creative endeavour of musicians.
“We need to find better ways of supporting creative workers, particularly in the music industry,” says Senator Hanson-Young. “We’ve been looking at what we can do in terms of tax incentives for live music events and venues and how we can really ensure that we stop those cancellations and actually feed some proper support into the sector.”
Moderator Nicholas Pickard of the Australasian Performing Rights Association mentioned the different political discussions around ChatGPT and OpenAI, which use different sources on the Internet.
“Some of the basic principles of the EU AI Act are that there is now a requirement for platforms to be transparent about the data they’re using to build their platforms,” says Pickard. “The US is moving somewhat in that direction as the EU. They’ve got the California AI Transparency Act, which is currently being debated at various committee levels in Congress.”
Both Pickard and APRA CEO Dean Ormston mentioned that a survey was sent to 4,500 members about AI. The results showed that 30% have adopted AI to help with music creation, another 30% rarely use the technology, and another 30% refuse to use the programmes.
From a copyright standpoint, Australian Recording Industry Association representative Annabelle Herd further elaborated on the changing laws around AI programmes.
“If you’re making reproductions to train generative AI systems and you don’t have permission, there’s no exception that allows you to do that. And that’s great. The problem we have is that who’s to know what is being used now. What if your work is being used, whether it’s publishers or labels or artists? Who’s to know what’s being used?”
Herd and Ormston further elaborated that mechanisms need to be used where people who are using copyright materials to train and generative AI systems will notify what is being used.
From a balanced perspective, Henry Compton of FUGA mentioned how his clients use AI in different ways. Some examples include programmes like BandLab or SONG Start-up and content ID recognition to prevent fraud and ensure that the rights holders are compensated.
“I’m hearing some really common themes in terms of consent and in terms of the right person getting paid from a very finite pool of revenue that we all deal with,” says Compton. “We need to ensure that creators do have that voice about whether their intellectual property is being used in these models.”
“I think it’s important that we don’t conflate AI and fraud as the same conversation.”
Another point made by Emmy Feldman of OTM Publishing was the convenience around AI usage, especially when it comes to backing vocals. However, in personal experiences, Feldman mentioned that recordings with AI usage were often not selected compared to singing from humans.
When an audience member asked what APRA was doing in the scenario where musicians have their songs sampled without being paid, Ormston mentioned that music recognition technology is slowly being incorporated to identify where a song is played, even without permission.
Similar to the software mentioned in Berlin, this would be an example of AI technology being utilised to track and notify artists and labels when their songs are being played at clubs, festivals, or livestream platforms.
In a positive perspective of AI in the music industry, Ellie Rumbold of Partisan Records described a client using AI technology for a creative purpose.
In the example, Partisan Records and the British band IDLES collaborated with Coldplay to create a music video for their song Grace by using a programme to replace Chris Martin’s lip movements in the Yellow music and “sing” the new lyrics.
Rumbold described the collaborative process, with Martin being recorded to sing the song by ten cameras, which captured five million iterations. The contract between Coldplay and the production company stated that the AI recordings couldn’t be used to train data for any machine learning systems and models.
“I think this is a really good example of where AI is really exciting, and we used it in a way to elevate and bring an artist’s vision to life,” says Rumbold. “But I think the important takeaway here is that we sought permission at every step of the way.”
