Our journey into the world of AI continues. We have previously talked about opportunities, challenges and also went hands-on (read our latest blogpost and update). This session takes on the ethical issues. Hendrik Vincent Koops gave a short master class around the topic of responsible AI: the ethical and social implications of AI and how we as a (media) music industry can strive to use this technology responsibly.
In this panel: Hendrik Vincent Koops (Senior Machine Learning & Audio Scientist, Universal Music, NL), Ivo Witteveen (Editor, Private Kitchen, NL), Than van Nispen (Researcher, lecturer, biologist, music designer, HKU, NL), and Rens Machielse (moderator, composer, producer, Private Kitchen / MiMM, NL).
AI and UMG
Koops explains what kind of role AI plays within Universal. Universal Music Group (UMG) is at the forefront of integrating responsible AI practices into their operations. ‘From artist involvement in AI processes to ensuring the ethical use of data, UMG is committed to protecting artists’ rights while leveraging AI to enhance creativity and business efficiency.’
‘UMG is committed to protecting artists’ rights while leveraging AI to enhance creativity and business efficiency.’ – Hendrik Vincent Koops
He emphasizes the importance of transparency, artist consent, and the protection of intellectual property in AI applications. UMG’s approach is guided by initiatives like the Human Artistry Campaign and Fairly Trained, ensuring that AI is used ethically and responsibly within the music industry.
Human-in-the-loop approach
UMG also develops AI tools for content protection, quality assurance, and playlist generation, always maintaining a human-in-the-loop approach to prevent automated systems from overshadowing the creative and ethical standards that are central to their mission. By partnering with AI companies like Endel and BandLab, UMG continues to explore innovative ways to integrate AI into music production and distribution.’
A case study: The Beatles
‘A great example is the one that has probably had the most impact in the last couple of months. You’ve probably seen this, the first new Beatles track. Basically, it couldn’t have been made without AI.
‘So, this was a track that was recorded by John Lennon on his piano on a cassette. The band actually tried to create a song out of this in the mid-90s but they couldn’t do it because the recording was so bad. When Peter Jackson was creating this documentary, he created source separation tools. These are AI tools where you can feed in like a whole mix of a track and it will do source separation of tracks; the individual guitar track, drums, keys and voice for example. In the end, Paul McCartney recorded the bass track, Ringo Starr drummed and Giles Martin created the string arrangement, and they used the original voice of John Lennon, which I think is an incredible example of where this kind of AI can be used to get new content without generating anything.’
Could anybody do that? Let the experts at Private Kitchen advise you.
‘I think when you create these models, you need some kind of musical background, especially working with the elements of music at the core. Of course, you would need data to generate music, but that’s more of a legal issue, in terms of being able to use the elements.’
BMIM Special best combination of theory and practice
In addition to this inspiring panel, the AI Composition jury of Private Kitchen announced BMIM Special as the best combination of theory and practice! And Manel Fornells (26) won the best composition with AI. Read more about Private Kitchen’s AI Composition Competition.
Pictures by Birgit Bijl
Text by Meike Jentjes