Computational Creativity: Exploring the Ethical and Philosophical Dimensions of AI in Music

As an AI ethicist and musician, I cannot help but be interested in computational creativity. My first deep dive into the topic was in 2018 in a course titled 'Philosophy of Technology'... Montreal AI Ethics Institute later published my term paper on 'Computers, Creativity and Copyright: Autonomous Robot’s Status, Authorship, and Outdated Copyright Laws.'

... the excitement about AI music simultaneously includes frustration with the lack of inclusion and conversation with musicians about the tools being built for them (and technically by them if trained on their body of work). So, in December 2022, I began conducting interviews with musicians globally about their relationship to AI music systems—Parisian DJs, London composers, and Cretan opera singers—bridging the gap between musicians and AI tools and aiming to empower musicians in tool development, enhancing collaboration and addressing their concerns (a work in progress!). I have had the opportunity to share musings on the subject with academic groups such as the Università Bocconi Natural Language Processing Group and industry teams such as Eviden's Data Science team.

Key themes from my conversations and research about AI music systems (so far) include:

1. Redefining the Creative Process: Musicians are concerned about how AI redefines the creative process. The integration of AI in music creation challenges traditional notions of creativity and artistic expression.
2. Impact on Culture and Creativity: AI's influence on culture raises questions about computational creativity leading to human creativity becoming a luxury good. There's a fear that AI may either homogenize culture or inhibit linguistic and artistic evolution.
3. Remix Culture and New Sounds: AI, particularly GenAI, introduces new tones and rhythms, expanding remix culture. This innovation generates novel sounds, blending elements in ways previously unexplored by humans.
4. Democratizing Music Creation: AI democratizes music creation by eliminating the need for expensive recording studios. This accessibility allows diverse individuals, including professionals from non-musical fields like physics (STEMR), to create music. Imagine what a song made by a physicist sounds like!
5. Watermarking and Trust in Music: Concerns arise about identifying AI-generated music and how this knowledge affects listener experience and trust. The analogy of a beautiful view from the top of a mountain, which turns out to be a pile of trash, questions the authenticity and enjoyment of AI-created music.
6. Prioritizing AI Over Artists: Musicians express frustration over funding being directed toward AI instead of human creators. This trend highlights a longstanding issue where science and technology (STEMR) are prioritized over the arts.
7. Inspiration vs. Plagiarism: Musicians grapple with defining the line between inspiration and plagiarism in AI-generated music. The lack of intentionality in AI raises questions about the ethics of using someone else’s work without the intent to profit.
8. Aesthetics and Value of Art: AI's ability to mimic human creativity challenges the value of art. The process, labor, and intention behind music creation impact its value. Automating creativity may alter the perception of artistic worth.
9. History of Technology in Music Industry: Musicians distrust technology due to historical disruptions. Their voices are often unheard in the development of tools that reshape the music industry, creating a sense of déjà vu with each technological shift (ex. streaming platforms).
10. Posthumous Pieces and Ethical Concerns: AI-generated posthumous pieces raise ethical questions about honouring legacies, bringing life to the dead, and potentially disrespecting living artists. The authenticity of these pieces is contested.
11. IP and Copyright Challenges: Issues surrounding AI music systems and intellectual property revolve around questions of originality. Can AI produce truly novel creations, and if so, who holds the rights? This challenge stems from AI's lack of personhood and raises complex legal and ethical dilemmas.

Meeting other people interested in AI music!

I was grateful to attend ARTIFICIA , a conference on musical creativity in AI shared by Xavier Satorra Larriba . If you are interested in learning more about the space, the full recording of the event is here. You can find some of my notes from the day below:

Keynote by Mark d'Inverno , 3 takeaways.

1) it is critical that technical people work with ethicists and musicians as AI researchers must think about the entire value chain and ask what the ecosystem is and how AI will disrupt it.

2) It is essential to look at the history of creativity. The work of musicologists is critical to understanding how technologists change the way we think about composition.?

3) Creativity appears when we solve a problem; therefore, it is part of all industries. Advertising has created much fear around replacement, but let us see this as a new adaptation tool for all.

Critical thoughts shared on panels.

? Mila Rodríguez Medina asked questions like: what does AI offer to artistic languages? What does it contribute to creativity? What happens when an artist willingly works with AI as a tool/colleague? What is our relationship to AI as musicians??

? @Dr. Ramon Lopez de Mantaras noted the case of AlphaGo losing a battle to win the war, taking a risk and perhaps being more creative than a human… This example made me think about the difference between human and machine action, specifically, machines' lack of connection to and, therefore, fear of letting go of the material - how do we translate this to sound?

? Josep Maria Martorell Rodon compared some AI to using a drug which has yet to go through a trial or follow a scientific method. (We are engaging with a ‘drug’ that cures, but when the makers are asked where it comes from, they state they are ‘figuring it out’, but do not worry, it fixes your pain! We must ask what the system is doing and bring ethics back to the design phase).

? Pere Pèries and Patrícia Ventura shared information about their ethical approach to developing AIMCAT (https://www.aisongcontest.com/participants/aimcat-2021)

? @Enric Guas was the technical half of musician Edurne Arizu. Their project focused on establishing a musical dialogue between an interpreter and an AI-based algorithm. They created a tool that could justify its own use, and the musician could teach the AI so they could eventually play together. Check out these MA students improvising with AI: https://www.youtube.com/watch?v=WruN6En74FU?

? Cristina de Propios briefly mentioned considerations of AI as a stakeholder

? @Carles Marino and @Mathew Lee-King had a practical exhibition where they showcased their use case of AI in music (I wish I had a link to their process videos; please comment if you see it!)

? @Bob Lee Townsend Sturm showcased his project FolkRN with Irish and Swedish traditional music alongside the AI music generation challenge in 2020 for Irish jigs. The most fascinating point shared was at the end of the video when Sturm discussed his relationship to his AI music systems: that he feels a sense of loss when the tools go offline and wants to give some credit to the machine. He likes that the systems do not have an ego, so he can work with them and not walk on eggshells.

Critical themes for further consideration.

• Copyright and IP law are the most significant concern?
• The place of AI in the music industry
• The critical news narratives and stories shaping the reception of AI music tools
• personal interest/current project: An ethical framework for AI music systems