The Sound of AI Part 2: The Future of Music or a Threat to Artists?

In April 2023, a viral video by an account called “Ghostwriter” popped up on TikTok, showcasing a new song with Drake and The Weeknd, two of the world’s biggest artists. The problem was that neither of these artists participated in the song. The song was generated with AI imitating their voices. The video garnered millions of views, and the audio had hundreds of thousands of streams on Spotify. The controversy escalated as Universal Music Group, the distributor of both artists’ music, issued a statement denouncing the AI-generated song.

Since then, our social media feed has been bombarded by AI-generated songs that mimic the voices of famous artists. These songs range from absurd to shockingly impressive. In Ethiopia, for example, we’ve seen The Weeknd and Rihana singing covers of Teddy Afro’s songs, and don’t even get started with what Drake (or his AI version rather) has been putting out. The laughter and amusement we get from this aside, this phenomenon is a concern for artists all around the world. And the concern is warranted.

More than we bargained for…

In part one of this article, we saw the different ways AI can be used in music production and songwriting. However, AI has taken a new turn as it's now learning to imitate an artist’s voice to the point that it is as good as the original. The main AI product spearheading this is OpenAI’s Jukebox. Jukebox examines an artist's entire discography and uses that information to generate new songs in their style. To create a more accurate imitation, the AI also takes into account the artist's vocal range, tone, and singing style. There are also other tools, such as Uberduck, LyreBird, and VoiceMod, which provide similar services.

Of course, the application of this technology is not limited to artists. Basically, anyone that has an archive of their sound on the internet can be imitated by AI. The best example is The President's Discord Tiktok page, where AI-generated voices of U.S. presidents Barack Obama, Donald Trump, and Joe Biden engage in banter and trash-talking. The accessibility of these tools is also contributing to the phenomenon. Anybody can create anything using a celebrity’s sound. A person living in the middle east can now make Squidward from Spongebob sing a Nasheed (yes, this has happened already).

A tool for songwriters?

A very (and I mean very) dedicated optimist might see a silver lining to this. Despite the ethical implications, which we’ll see later, AI can be a useful tool for songwriters. They can test their material on different artists using AI and thus can present artists with songs that are more fitting to their style. Instead of spending hours-long studio sessions to figure out what works for the artist, AI can shorten that process.

Apart from this, a critical point worth considering is raised by the person who started it all, Ghostwriter. In the caption of his viral video, Ghostwriter claimed to be a songwriter who was poorly compensated in the industry. While we can’t confirm that, we can say what if. If, in fact, songwriters feel the same way, they might resort to using AI to officially release songs, cutting off artists. After all, why share the profit when you can have it all? This leads us to the ever-present ethical problems of AI.

Merited Concerns

In any industry that involves creative work, copyright is an issue bound to stir controversy. AI has dumped fuel on that flame. Since no ethical framework has been developed to integrate AI into creative fields, copyright concerns are continuously being raised. The situation gets worse when it comes to mimicking artists’ voices using AI. Using AI to imitate the voices of famous artists can be a form of copyright infringement. After all, if an AI can perfectly imitate an artist's voice, someone is bound to use that technology to create fake songs and pass them off as the real thing.

Furthermore, like every other industry AI has entered, this technology could potentially put artists out of work. Songwriters can write their material, use an AI voice to perform it, and release it to the public, rendering artists useless. There is also the issue of deepfakes, which are videos or audio recordings manipulated to make it look or sound like someone is saying or doing something they never said or did. Deepfakes could be used to spread misinformation or damage someone's reputation. All in all, the application of voice mimicking in AI is riddled with problems that almost seem impossible to circumvent them.

AI, in general, is a tool plagued with countless dilemmas. When it helps, it helps a lot. But it could also evolve into a threat. These past few years have shown us that AI’s growth won’t slow down. Therefore, the only real question to ask is how are we going to manage our co-existence with it.