Deharmonizing or Dehumanizing: A.I. in Music

 Aidan Munoz 

Professor Horgan 

Seminar in Science Writing 

4 May 2024 

Deharmonizing or Dehumanizing: A.I. in Music 

With each passing day, humankind dives further and further into the innovation rabbit hole of artificial intelligence. Machines and programs are constantly getting smarter, more intuitive, and in all honesty, more humanlike (at least enough to pass an average human’s “Turing Test”). With all of these advancements being made in academic and scientific fields, such as training robots to do surgery, what does this mean for our entertainment and creative industries? Have the capabilities of A.I. advanced enough to become autonomous creative entities, or are they merely a tool that can be used and abused by those with the ability to do so? This brings up the point of my research: Can the music industry ethically use artificial intelligence when creating music? Now, as a preface, I would like to submit my answer to the question as yes, however, just because it can be used ethically, does not mean it will be. 

My research is split up into two major parts. The first being research done on what A.I. music is, recent major advancements in the industry, and how it can be put into use. The second being ethics, morals, and legal implications with AI music. Perhaps the biggest ethical implications to be focused on should be the future of posthumous albums, the current underdevelopment of AI copyright laws, and how creative intellectual property can become stolen work without payment or recognition moving forward. 

First and foremost, what is A.I. music? A.I. technologies are revolutionizing music creation by providing new tools and platforms for musicians and composers. They can generate musical compositions, assist in songwriting, and even create personalized music experiences for listeners, expanding the creative possibilities in the industry. AI-powered recommendation systems and music streaming platforms utilize machine learning algorithms to analyze user preferences and behavior, offering personalized playlists and discovery features. This enhances the listening experience for consumers, leading to increased engagement and satisfaction. A very recent and major proponent of this market is Udio (released in 2024), a generative A.I. model that creates music based on prompts, created by a team of former researchers for Google DeepMind. In partnership with Microsoft comes another major player, Suno AI, which is a very similar company, and was initially released in December of 2023. Suno also works in tandem with OpenAI, using their ChatGPT AI to generate the lyrics for their songs based on the inserted prompt. Now the musically creative aspects of these companies feel endless and are available to the public on their free websites. Based on a prompt describing genre and/or a description of what the song should be about, the A.I. systems provide the user with a completely brand-new song in, at least in my experience, less than a minute. The website provides you with all of the lyrics written out, the audio, A.I. generated cover art, and the option to download the audio and/or a video that shows the lyrics as they are performed in the song. Once generated, you also have the option to “remix” by adjusting the prompt or changing the lyrics manually, or even to “extend” the piece and ask to add/generate additional parts to the song. According to a report by Rolling Stone, “Suno says it’s in communication with the major labels, and professes respect for artists and intellectual property — its tool won’t allow you to request any specific artists’ styles in your prompts, and doesn’t use real artists’ voices”, which helps navigate very obvious copyright issues (Hiatt). Although it is unclear if they have run their algorithms on countless hours of copyrighted music, it doesn’t really change much when their learning model and algorithm is still trained on thousands of hours of music that is free for commercial use and/or just not copyrighted in general. I suggest you try it out for yourself and see how effective it is at replicating human art. 

As a result of these advancements, morally, the topic of posthumous albums is one that cannot be ignored. Artists passing away young is a sensitive subject that has been quite common in the past few decades, just as common as their creative works being released after they pass whether they wanted them to be released or not. If labels and/or artists’ family members already have an inclination to drop their creative works even if incomplete, what is to stop them from now creating entirely new artificial and synthetic albums using AI algorithms? If an artist’s entire style, voice, and character can be artificially replicated, can they not live on forever? Although a lot of labels, companies, and even family members may use these new resources as a larger money grab or to keep the career of their artist going, there are some positive ways things can be done. For example, a nonprofit called “Over The Bridge” employed Google’s “Magenta Program”, which is an AI algorithm with the ability to simulate “new” songs after analyzing an artists’ past works to create new works from Kurt Cobain and Amy Winehouse (currently deceased artists). The purpose of this was to bring about attention and awareness to mental health and recovery by showing what kinds of work we lost after the passing of these blossoming young creators. On the tech side of things, Rolling Stone explained it as “after examining each artist’s note choices, rhythmic quirks, and preferences for harmony in the MIDI file, the computer creates new music that the staff could pore over to pick the best moments” (Paul). Is it right to grab a bunch of songs from a deceased artist, throw features from another artist they didn’t want to work with on them and put it out because they can no longer argue against it? Is it justifiable to use AI to finish a nearly finished project so that it doesn’t “go to waste”? In all honesty, there is no one answer. Everything is case by case. However, an artist’s wishes to complete a project or leave it unfinished without their creative input should always be respected, as it is their intellectual property. 

Consequently, because of the work that can be done without the original artist being present or applying their individuality to a piece of art, there are ethical issues and concerns in terms of the law that must be addressed. As James Sammataro and Nicholas Saady, two very high-profile and experienced attorneys in the music industry, expressed to Billboard, “the quality and efficacy of AI is ‘evolving faster than the courts can evaluate how laws apply to it’” (Sammataro and Saady). The United States’s Copyright Act protects exclusive rights for authors and declares that an author must, therefore, be human. However, this brings about the issue of trying to discover and pinpoint how much human intervention and influence on an AI generating music will then make the process copyrightable. As, in the meantime, artists will continue to have their intellectual property stolen from without having their rights protected, as some U.S. courts draw a thin line deciding that AI artwork cannot be a copyrightable work within itself as it doesn’t have enough human intervention to declare it having a human author (Sammataro and Saady). Due to the fact that AI art creation is still such a relatively new topic, laws have not fully been formed to decide how to move forward with them, but there will be an emphasis on human interference and direct change on it before the conversation can be had. Interestingly enough, Suno and Udio claim that when free account users use their generation, the companies retain all ownership of the songs, including the artwork. However, paying subscribers own the songs and the artwork. Does this “transference of ownership” entail copyright ownership? Because if so, then what happens if I use “my song” for commercial use? Perhaps that is partially why these more recent generative tools give you the option to remix and extend their generated work, hoping that if you change certain factors of the song, you will now become the author and own the copyright.  

Unfortunately, the industry won’t know the limitations until it is a larger problem and brought to the courts or addressed in award shows. On June 16th, 2023, in an interview with the Recording Academy CEO (the entity behind the Grammy awards), Harvey Mason Jr., he declared, 

“At this point, we are going to allow AI music and content to be submitted, but the GRAMMYs will only be allowed to go to human creators who have contributed creatively in the appropriate categories. If there's an AI voice singing the song or AI instrumentation, we'll consider it. But in a songwriting-based category, it has to have been written mostly by a human. Same goes for performance categories – only a human performer can be considered for a GRAMMY. (Enos)” 

Now, this still feels very vague, but that at least provides a general strucutre of how things will proceed. Of course, there are the nitty gritty details of what actually constitutes as enough human intervention/contribution, but answer may come with time. Proceedingly, how will this affect artists’ numbers, fanbase, and creative process? We will just have to wait and see what happens in the near future. 

Overall, with all of the ethical implications that AI music brings about, why does it matter? AI music isn’t all bad and perhaps instead of fearing and/or loathing its existence in the realm of creativity, perhaps embracing it as a new form of creativity and inspiration is the way to go. As Sarah Mackenzie, a guest writer for the CBC News put it: 

“New technologies have pushed art forward, facilitating the birth of musical genres, from hip hop to techno. As we head toward a future in which AI will inevitably play a more prominent role, the technology's capabilities and intentions need to be deconstructed and understood. We need to build collective understandings and belief systems around the role technology plays in the way we create. (Mackenzie)” 

As companies like Google continue to create AI systems like “MusicLM” that have been trained on datasets of hundreds of thousands of hours of music to generate coherent songs with complex descriptions, the problem becomes less clear to pinpoint, yet easier to enjoy (Bhattacharyya). This way, AI can now be utilized as a tool for inspiration for real artists and creators looking for a new spark. However, for this to be the case, we should have a justified reason for it. If it is for a lack of or desire for more music, then it would be more beneficial to support performers and musicians, and fund music programs. Back to the point of the technology, rather than assuming AI will eventually be harmful and damaging the more it evolves and eventually gains this looming sentience we fear, we should embrace the capacity of these algorithms to enhance our abilities or communicate and express ourselves with one another through powerful outlets (Sato and McKinney). Other than the fact that it gains all of its “inspiration” from the work of other artists, AI opens a door of newfound effectiveness and efficiency for artists to enhance their work and creative processes and actually benefit the experience of the listener, when done the right way. 

Taking all of this into account, I decided to speak with a graduate of both Berklee College of Music and Columbia University, a former Assistant Arts Professor at NYU, an accomplished audio engineer and musician, and current lecturer at Stevens Institute of Technology, Rosana Cabán. We began with the topic of ethics, landing on the same conclusion that there have always been ethical issues within the music industry, and unless there is lobbying against it, the problems will continue. There are well-known issues within the industry of unfair contracts and poor working conditions, so if the major labels find a way to cut corners and utilize A.I. for monetary interest, those will be the routes they take as they don’t work to benefit artists, they work to benefit the company. When asked about if A.I. would be able to compete with the creative innovation of humans, she responded,  

“Artificial intelligence is just replicating information that is fed to it. It can analyze song structure and chords and create meaningless lyrics with vague references to heartbreak or partying, which is already happening in pop music . . . However, there will always be the outliers creating innovative music (like Rosalia, Arca, Bjork, or Kaytranada) who play with the ‘rules’ of structure or mixing frequencies.” 

There is so much innovation that the human mind is able to generate through trial and error and our creative processes that make our creations special. A.I. doesn’t understand the concept of how a song can make you feel, rather than how certain patterns of melodies, rhythms, and colors are deemed “appropriate” for their tasks. It can replicate and copy patterns all it likes, but for it to create something truly new and unique, it would need the guiding hand of a human. In general, her overall view on the subject was, “A.I. may help disabled people make arrangements, or people who lack resources to expensive sound libraries, etc. I can see it being used for good if integrated with a human perspective. I don't see it making better music than a human - ever.” 

In essence, I do believe the music industry can ethically use artificial intelligence when creating music, however, I do not believe that it will. Wherever major companies will be able to cut corners, they will. The draw in of numbers from listens and streams will more than likely influence them to drop even more posthumous projects from deceased artists due to the fact that with the number of vocal tracks a label would have from any of their artists, they could create a nearly perfect audible replica of their performer and graft the voice cover on a hired singer performing the right notes. Ethically, there are still countless issues that need to be addressed as we move forward with how anyone can use these tools and do anything digitally to make it look and sound more real than you could imagine. As those copyright laws get sorted out, we may see a better, more creative future, or we may experience a drought of out of the box creativity as the laws become so strict, they bind human artists themselves from using these newer tools. Outside of that, AI can be used to truly benefit art and its creators in so many forms of creative enhancement and inspiration that perhaps we just need to stop being afraid of, and instead decide for ourselves to utilize these tools. Regardless, no matter how these things play out, it seems that AI music may be the key to immortal human creativity or eternal creative suffering. 

 

Works Cited 

Bhattacharyya, Andy. “Music Therapy: Ethics in Generative AI for the Radio and Music Industry.” Medium, UX Planet, 15 Apr. 2023, https://uxplanet.org/ethics-in-generative-ai-for-the-radio-jockeys-48925fbc77b6. 

Enos, Morgan. “Recording Academy CEO Harvey Mason Jr. On How the New Awards Rules and Guidelines Will Make the 2024 GRAMMYs More Fair, Transparent & Accurate | GRAMMY.com.” Grammy.com, 16 June 2023, www.grammy.com/news/recording-academy-ceo-harvey-mason-jr-discusses-2024-grammys-new-rules-guidelines. 

Hiatt, Brian. “A ChatGPT for Music Is Here. Inside Suno, the Start-up Changing Everything.” Rolling Stone, 17 Mar. 2024, www.rollingstone.com/music/music-features/suno-ai-chatgpt-for-music-1234982307/. 

Mackenzie, Sarah. “Artificial Intelligence Is Shaping the Future of Music - but at What Cost? | CBC Music.” CBC News, CBC/Radio Canada, 24 Mar. 2021, https://www.cbc.ca/music/artificial-intelligence-is-shaping-the-future-of-music-but-at-what-cost-1.5952955.  

Paul, Andrew. “An AI Composed Posthumous Songs for Kurt Cobain and Amy Winehouse.” Input, Input, 5 Apr. 2021, https://www.inverse.com/input/culture/ai-composed-posthumous-songs-for-the-likes-of-kurt-cobain-amy-winehouse.  

Sammataro, James and Saady, Nicholas, et al. “How Much Humanity Will AI-Generated Songs Need to Be Copyrightable? (Guest Column).” Billboard, 27 Jan. 2023, https://www.billboard.com/pro/ai-generated-music-songs-copyright-legal-questions-ownership/. 

Sato, Maki, and McKinney, Jonathan. “The Enactive and Interactive Dimensions of AI: Ingenuity and Imagination Through the Lens of Art and Music.” MIT Press, MIT Press, 4 Aug. 2022, https://direct.mit.edu/artl/article/28/3/310/112448/The-Enactive-and-Interactive-Dimensions-of-AI. 

“Suno.” Suno.com, suno.com/. 

“Udio | AI Music Generator - Official Website.” Udio, www.udio.com/.