AI Music is Here: Get With It
ART BY AUDREY NJO
The use of generative artificial intelligence tools has remained a nonstop conversation since their rapid popularization in the past year. Most commonly illustrated by generative chatboxes like ChatGPT and Google Bard, the world has been marveling at AI’s ability to complete conventional tasks with unprecedented efficiency. These tools effortlessly generate responses according to the user’s request, within seconds ranging from Christmas gift ideas to sports predictions to five-page essays to even computer programs. While the immediate benefits of these tools are undeniable, the unsettled ethics of their usage have accompanied generative AI since its inception.
Currently, there is no clear definition of “fair use” for generative AI, and the lack of comprehensive understanding of its usage has become drastically disproportionate to its accessibility. Colleges and universities were alarmed when the usage of these tools to complete assignments became widespread among student bodies, with one study reporting that 56 percent of 1,000 surveyed students have used AI on assignments or exams. AI’s capabilities have also caused reasonable fear of job displacement, as some believe it has the potential to replace 800 million jobs worldwide by 2030. Most recently, generative AI has expanded its cloud of controversy towards the music industry, once more sparking countless debates.
Today, generative AI companies are offering services for users to create music in a variety of ways. Applications such as Soundraw are trained on principles of music theory, allowing users to generate compositions by describing their ideal vision in terms of mood, tempo, melody, and more. On the other hand, tools like AIVA allow the usage of pre-created material to generate new content, advertising its ability to produce variations of existing songs. Some of the output may include rhythms and lyrics similar to existing music, whereas others fully adopt the style and vocal profiles of famous artists. Out of these general categories, the latter has garnered significant pushback from industry professionals, as it permits the unauthorized usage of copyrighted material.
On social media platforms, users are creating and posting imitation songs made with AI technology for profit without the artists’ consent. One particular account on TikTok called “@ghostwriter977” has attracted over 250,000 followers. Despite being flagged and banned by the platform, this account has managed to retain its popularity and continues to post the same type of content.
The account’s most recent post, which has received just short of 100,000 likes, features a song entitled “Whiplash” that clearly imitates popular hip-hop artist Travis Scott. The dark, eerie trap melody emulates Scott’s signature style, while the autotuned vocals and high-pitched ad-libs are virtually indistinguishable from the original artist. More comically, the song even features a beat switch and a similarly AI-generated verse from fellow rapper 21 Savage. The comment section captures mostly positive reviews. Some users are demanding the song to be uploaded to streaming platforms like Spotify, and others are calling for “@ghostwriter977” to be nominated for major awards such as the Grammys.
However, there are also voices questioning the ethics of this trend given that the song blatantly replicates artistic identities to attract attention. “In my opinion, AI could be a positive force for… production in music, however, I have seen time and time again it being used to create soulless works and [detering] some artists from a much harder, complete creative process,” said sophomore Connor Howe in a written statement to the Observer.
This divide on AI can be felt in the music industry itself, illustrated by Ed Newton-Rex’s resignation as the vice president of audio at Stability, a generative AI company. In his op-ed for Music Business Worldwide, he attributed this decision to the company’s stance that “training generative AI models on copyrighted works is ‘fair use.’’’ According to Newton-Rex, this viewpoint is widely shared across various pioneer companies in the generative AI industry. Though he maintains his confidence in the potential benefits of generative AI, Newton-Rex states he “can only support generative AI that doesn’t exploit creators by training models—which may replace them—on their work without permission.”
Luis Cárdenas, who works as a specialist of Creative Administration at Columbia Records, independently agrees with Newton Rex’s standpoint in his written statement to the Observer. When asked about the concept of fair use, he writes that “to consider such uses as ‘fair use,’ you must first ensure that the party (in this case, the recording artist) has agreed to their use of likeness (their voice) in such a way.” Unfortunately, this is often not the case, as seen in Puerto Rican artist Bad Bunny’s outraged denunciation of AI imitators in his WhatsApp channel, telling fans who enjoy these generated songs to not attend his upcoming tour.
Regulation may prove to be a difficult task, as the accessibility of generative AI has spread far beyond the professional level. “It’s challenging to prevent individuals from using copyrighted material for training AI models, especially when this material is publicly accessible online,” said Patrick Mauboussin, a computer science student at Georgetown University who has been working closely with generative AI since the GPT language model’s polarization.
Tufts Music Professor Paul Lehrman expresses a similar concern, “There are going to be many, many legal cases… I think we are in for a long struggle in which a lot of this will not be settled for a long time, if ever. As long as these algorithms are in operation, there’s going to be controversy over how they are used.”
Nevertheless, Mauboussin believes that artists’ role will not merely diminish as a result of generative AI’s competence, Mauboussin stated “These tools are better seen as assistants rather than replacements… highlighting the importance of human expertise in guiding AI to achieve desirable results.”
Regardless of these nuances, Cárdenas thinks the general public should keep an open mind about the potential of AI. Despite how it may be misused in instances at this current stage, he explained that “We ought to consider whether our fear of change is interfering with our ability to understand these advancements and how we can make better uses for them.”
In his eyes, drastic shifts that could come with the continuous development of generative AI are not unprecedented. “We can look at the last 20 years and analyze the impacts of Napster or how Apple/iTunes disrupted physical album sales with the convenience of digital purchases and musical libraries… All of these developments completely altered the traditional ways in which recording contracts, access to music libraries, or sales and charts were determined,” Cárdenas said.
Lehrman agreed, suggesting that the concept of AI music may not be as jarring as it sounds. “I think the whole idea of using machines to generate music is not a new one,” he explained, “It’s just that we have these new tools that are much more sophisticated than the previous ones.”
For now, the most pressing issue seems to be the protection of artists’ rights, especially with the ambiguity of AI ethics. “As to whether the average audience could spot AI music or even care, I don’t know,” said Lehrman.“The only people who’d care might be the artists whose work is being mined. If the artists may be properly compensated, it might be able to benefit everybody.”
However, he adds that “the vast majority of contracts favor the large publishers and labels behind the artists, as the artists rarely get to keep all of the rights to their music. If the companies decide to license out an artist’s music to AI companies, the artist may not be able to do anything about it.”
While the impact of these potential issues remains to be seen, the train of AI has already been set in motion with all its thrill and terror. As Cárdenas says, “AI has already entered the zeitgeist of the music space, so the future is pretty much now.”