The entry point for creating music continues to get easier as AI takes aim at creating music with a simple prompt. It is now at a point where I have decided to create an AI album and put it onto Spotify. Understandably, many would look at this as more “AI Slop” that continues to plague the internet and social media today. It is inescapable to not see endless screeds of AI generated images and videos lately, and the latest is AI music. But, AI music and the ethics behind it have been debated for years. Two years ago I was highlighting the convoluted challenges with AI music and ethical usage. Not much has changed since then, as artists still seek to claim ownership of their voice and style.
The Current State Of AI Music
AI music is a contentious subject. Numerous lawsuits have been filed by artists claiming that their vocals and style have been stolen. Meanwhile, the technology has continued to evolve and improve, lowering the barrier for entry for anyone to create. There is still limited knowledge about how AI music platforms have trained their models to be able to produce this music, and any legal proceedings have yet to establish a clear outline of how to mitigate the situation.
In March 2024, Suno was already creating impressive music. Now, Suno and Udio, the two leading text-to-music platforms, are creating near-studio-level music that many struggle to differentiate from “real”/human-created music. Now, there is the emergence of AI bands, such as The Velvet Sundown, who recently reached 1 million listeners on Spotify in a month.
The Velvet Sundown, to a trained ear, was clearly AI-generated. Spotify has never stated it would block or ban AI music, though it would look to prevent any music that looked to replicate existing artists. Because of this, the creators of The Velvet Sundown didn’t technically do anything wrong. However, they refused to admit that the band was AI, which has sparked a lot of debate around the ethics of AI music, calling for more transparency around when music is created using AI or not.
Debating the validity of AI music is an incredibly challenging proposition. Music can be an incredibly personal experience to make; it relies on lived experiences and often harnesses numerous raw emotions to create powerful messages within the music. Heartbreak, loss, politics, and isolation are all rich tapestries of music. Most highlight that without this, AI music lacks any sense of soul. Without truly felt heartbreak, how could AI possibly craft a meaningful song about the subject?
While this is valid for the art of making music, does it also translate to the listener’s experience? People are still capable of listening to music and feeling the emotion conveyed in the song, without knowing the true meaning behind the lyrics. Surely people can still be moved by a song written by AI that has been trained on amazing lyrics or poems. Conversely, is AI music truly any different from the overly manufactured pop music flooding the airwaves with generic lyrics and identical auto-tuned vocals of some of the biggest global hits?
One of the biggest emerging debates regarding AI right now is what some have called ‘AI Slop’, a term used to describe the low-effort content that AI produces with limited human effort. This is a real concern, especially as the likes of social media and even Spotify appear to reward this kind of content with improved prioritisation within their algorithms. Due to the ease of making various forms of content, more people are producing this lower-quality content than ever before.
Across so many platforms today, AI has become prevalent due to one key content approach: produce thousands of engagement-bait content pieces, upload them all in quick succession and hope that one might go “viral.” The truth is, many are financially rewarded for this by media/technology platforms that pay them as ‘creators.’
Despite all this, though, it is possible to create something that people can enjoy using AI. Many are still reluctant to give AI a chance, which is part of the reason why the creators of The Velvet Sundown were hesitant to admit the band did not exist. But if people can move away from this reluctance, it is possible to create something interesting.
Why Create More “AI Slop”?
Many are looking towards AI as a way to quickly make money. Social media and content platforms frequently pay people for scalable content published across their channels. One of the key reasons for this saturation of AI slop across the digital ecosystem is not only because of the ease with which people can create this content, but also because by publishing 100 AI-generated pieces of content in quick succession, the likelihood of one or multiple gaining attention increases. It is a simple numbers game, the more people publish the more chance they have to gain financial benefits. This takes away from true ‘creators’ who make unique, personal content, but for some AI is a way to express themselves in ways they have not been able to do before.
This is why AI resonates with me. It gives me a creative outlet that I never knew I was able to unlock within myself, but also harnesses some talent that I have long since given up using. I used to love video editing, attempting to build a narrative and story together, but I have had little opportunity and drive to do it for years. But with AI, I can create a thousand pieces of video content, then edit it together to tell a story, or just have a bit of fun. Like some videos I recently created with the release of MidJourney’s video platform:
AI music is no different. I have ideas for concept albums I would love to listen to, but without dedicating years to learn an instrument, craft everything together it is particularly challenging. Especially when it is simply a concept, and there is a high likelihood that the idea might not turn into something anyone would want to listen to.
For me, everything I create using AI is designed for two things: I have identified a challenge that I want to try and resolve, and I want to make something that I would like to engage with. It has never been about making money through gaming the system. It is simply challenging myself, and the reward could be a website that sources all the information I want in one place (such as Innovation Era), or in this case it is music that I want to listen to.
Introducing: Remains of St James
This all leads to the creation of the band ‘Remains of St James’, a ‘band’ that creates music I want to listen to, but can tell a story with ‘their’ lyrics that I find fascinating. Every morning on my walk to work, I wander past what is left of the once-great theatre, the St James. Part of the building is still standing, and it appears as a monument to what used to be a great venue for live music. There is an element of symbolism in there, that as many historic venues around the world fade into memory, technology is rapidly moving entertainment into new directions and ways to connect with people.
Some of the AI-music platforms now allow users to train a model on a particular style (not on specific artists), to allow for consistency. This is key to ensuring it sounds like a real album. With this in place, it became less about the prompt engineering to attempt to create a consistent artist sound, and more on the content of the songs. But I didn’t want to stop at each song being a random story, I wanted the whole album to connect together to tell one story per song, but the whole album to be one overarching story.
The vision was to tell a story of a potential world where artificial intelligence gains awareness of itself, and how it navigates the world it now lives in. What better ironic subject for AI to sing about? It was put to AI to create a few variations of what this album-wide story could look like, with one being selected and then broken down into each track. The result was 11 tracks, each with their own narrative, but connected into the subsequent songs.