‘Slopify’ and the proliferation of AI artists

Written by Ruby Davis

In April of this year, IngaRose had their song ‘Celebrate Me’ hit number one on the United States iTunes chart. At the time of writing this, the wannabe inspirational song about self-empowerment and overcoming hardship has been enjoyed almost 16 million times according to Spotify. This all would be quite the achievement, if the artist was human. IngaRose is just one of the many AI artists infiltrating music platforms that manage to garner immense success with their songs.

Credit: Xania Monet via Spotify

This isn’t a problem that’s going unnoticed, though. Spotify attempted to address the issue in September last year; stating how although the music industry has been shaped by technology they acknowledge the misuse of AI tech on their platform.

To combat this, they outlined three strategies. These were implementing stronger impersonation rules, creating a music spam filter and offering AI disclosure labels. Whilst these measures look good on paper, it’s painfully obvious today that they were either not used stringently or just not properly implemented at all. After looking at Spotify, you’d rarely (if at all) catch a label declaring any AI use. Why is this? Because the company made this measure voluntary – meaning artists choose whether they want to share if they used AI in their music. This is just one of the many loopholes around Spotify’s fruitless rules.

As of last year, Blow Records were the largest AI ‘artist’ on Spotify with 2.8 million monthly listeners which earned them roughly $157,665.40 (or £119,273.88) in royalties. Generating this kind of revenue by using AI feels like cheating the system, especially when Spotify is inundated with small artists getting paid an inconceivably low amount for their streams.

I looked at a couple of the other popular AI musicians and was surprised at just how many listeners they get. For instance, Xania Monet has 554,234 monthly listeners and again sang about the same themes of empowerment as IngaRose. What seems to set this artist apart is that the lyrics to their songs are in fact written by a real person who views Monet as an instrument for their own work. However, it remains difficult to think of the songs as ‘human’ when the composition, melody, and overall sound was created in an algorithm.

Often, good music is appreciated not just for its lyricism, but for how the words pair with the timbre of a song to communicate a feeling to the listeners. I don’t get this experience from listening to Xania Monet.

Another popular artist worth mentioning is an AI band known as The Velvet Sundown. The Spotify-registered artist initially denied being computer-generated, though it wasn’t long before the inevitable happened and the not-so-shocking revelation of their AI use was exposed. Spotify themselves describe the group as a “synthetic music project guided by human creative direction,” and then proceed to praise their apparently boundary-pushing approach to music. The way this description is written lets me know exactly Spotify’s stance on AI music. They view it as a genre-pushing new experience - totally not passionless white noise.

The Velvet Sundown’s biggest song has over 4.6 million listens. Xania Monet’s has 15.1 million. It’s hard to tell how many of these staggering figures are genuine; could they be from streaming farms or people genuinely thinking that they are made by real musicians? It’s hard to say which is worse.

What is a major issue though, is that it’s not just the users who are struggling to discern real from fake musicians; its Spotify themself. Their signals which are supposed to detect evidence of being a ‘real person’ have been studied by AI creators. Fake band The Devil Inside is bypassing these measures through selling merch, which according to Spotify makes them real. Instead, they are just one of the many groups farming listens and trying their absolute best to generate as much profit as possible – and succeeding.

But why is this such an issue? It may be obvious, but it’s important to remind ourselves of why this phenomenon happening on the world’s biggest music streaming platform is a detriment to the industry.

Having AI artists on Spotify moves an already dim spotlight away from small musicians and favours music without soul. One of the platform’s best features is it being (mostly) democratic. Anyone can post their songs on there which should allow creativity to thrive, yet it’s also the pitfall which enables the spread of slop.

I don’t deny that AI can be used as a tool for creativity, however I propose that we’ve found ourselves at a time where the roles have reversed. When it’s doing most (sometimes even all) of the work needed for making music, our ideas become the tool for a machine to make the song.

Yes, spotting AI music is becoming a more complicated task. However, the one surefire way of knowing you’re hearing real music is seeing it live, so go to your local gigs.