When is AI Music Not Large Language Mediocrity?

We are finally in the world foretold in those 20th-century sci-fi novels and films. Or at least in the earliest version that’s come to reckon with a discordant reality over the span of 5 years. Whereas the word ‘AI’ was once the exclusive preserve of nerds and hardcore tech enthusiasts, today, companies devote hundreds of millions of dollars to Claude token budgets. Bizarre art styles and linguistic forms have emerged from this techno-humanoid creation pool. Even now, as you read this, generative artificial intelligence (AI) probably plays a greater role in your day-to-day than loved ones. Complaints about AI proliferation—including an encyclical from none other than the Pope himself—are but sermons to the void. 

Of course, this also means art is affected; an outdated conversation in an age where AI film festivals exist. Creators seek to be ahead of the curve, as with all innovation from time immemorial. And so across different media, more people are expressing themselves via prompts. 

The music industry has been substantially impacted in this regard. Deezer, a streaming platform, reported on 20 April, 2026, that roughly 44% of daily uploads (75,000 tracks) were AI-generated, totalling more than 2 million a month. This figure was about 10,000 as of January 2025. AI songs aren’t springing up from thin air, either. Their ether is everything else that has existed prior. 

When the Recording Industry Association of America (RIAA) filed a lawsuit against AI music generation startups Suno and Udio on June 24, 2024, for training their models on copyrighted music, it came as no surprise to most that Suno’s response was to concede to the claim. In March, Billboard Pro reported that at least 7 million songs were generated on Suno each day as of November 2025. Certain experts consider Suno to be as disruptive as, if not more than, Spotify and Napster were to the music industry. And one need not have seen Justin Timberlake as Napster co-founder, Sean Parker, in Aaron Sorkin’s critically acclaimed 2010 film, The Social Network, to grasp the gravity of such a comparison. Music has changed, possibly forever. And over the next few years, everyone in the ecosystem has to figure out their role as we herald this new age of sonics. 

The trend of globalisation has meant that Nigeria isn’t left out of the fray, a description that can be interpreted either way. The biggest AI music moment, so far, took place in November 2025, after an AI choir called Urban Chords’ cover of singer Fave’s Intentions went viral on social media. Urban Chords’ version lifted Fave’s original vocal stems and converted them into a new, choral arrangement without her permission, as this Techpoint piece explains. Amidst the outrage, Fave released a remix of the record with the anonymous Urban Chords, opting for a longer shelf-life as opposed to taking down the cover completely. 

Singer Dammy Krane’s recent popularity is also tied to generative AI music. In what can best be described as desperate clout-hunting, Krane has released AI versions of multiple Nigerian projects in the past 2 years, mostly on the same day as the original projects. Krane’s 2026 run includes rip-offs of Omah Lay’s Clarity of Mind, Sarz and BNXN’s The Game Needs Us, Asake’s M$NEY, Kanye West’s BULLY, and Wizkid and Asake’s REAL Vol. 1, not accounting for singles. 

When a video for  Peter ‘Mr P’ Okoye, one half of the defunct mega-duo PSquare’s, latest single, I Can’t Look Away, went viral on X on 6 June, 2026, certain users pointed out that the song was likely a product of generative music AI. Producer and composer The Beatoven indicted Suno for the song’s writing and arrangement. Singer-songwriter Mizzle commended what he felt was a great, albeit undeniable, use of AI. Most reactions were positive, drawing comparisons to older PSquare singles. Davido even co-signed the song the following day. The implication here is that listeners care less about the process and more about music arriving on their DSPs. Ethics are an afterthought; an aside for a populace passionately fanatic about music. 

And the charts reflect that most people just do not care. A recent article by culture journalist Idowu Abdulmuqsit revealed that the year’s biggest song is entirely AI-generated. Let Me Be by AI group, The Second Voice, has surpassed 200 million streams globally. According to Abdulmuqsit, whose analysis was based on data from TurnTable Charts, Let Me Be’s numbers are the highest among African acts this year. 2026 hasn’t been a quiet year across the continent by even the most biased metrics. A-list artists have released eagerly anticipated LPs and surprise smash EPs. Singles for forthcoming projects have accompanied coruscating rollouts. Major labels, home and abroad, are pushing their artists with all the verve typically reserved for new acts. And yet, the biggest performing song isn’t even wholly human. Abuja, we have a problem!

In literature, AI has found its way to one of the world’s most prestigious literary fiction prizes. In film, more and more filmmakers are surrendering to the onslaught of the magic prompt. Beautiful visual art styles are fed to the machine and spat out with distinct sensibilities that might fool even the most astute curator. Music is simply walking in the path of its expressionist siblings. Which means that rage as we must, the inevitable rise of such art arrives all the same, strengthened in a world with diminishing taste in quality art and preference for shallow, shiny toys. 

If there are ethical ways to create music with generative AI, then we must find them. Knee-jerk catch all responses only suffice on social media forums where virtue-signalling is absolute currency. In the status quo, much else drives the adoption of AI music generation.  

Humans as beings of expression will always seek new ways to bring their ideas and emotions to life, even when the long-term effects are damaging. Many lack these creative skills. Many are not willing to put in the work required to develop these creative skills. Many lack access to proper learning experiences, limited by geography. Many lack exposure to realise they are capable of producing art in said way if only they tried. In the presence of these stumbling blocks, generative AI music becomes a mallet. A way out of the creative blockade. Hence why the majority using these tools aren’t even within the industry to begin with. Just regular folks ‘having fun’ with other people’s intellectual property; posting remixes and new tracks on their socials for external validation and a sense of accomplishment. 

For anyone needing a reminder who the real enemies are, consider that the big 3 (Warner Music Group, Universal Music Group, and Sony Music Entertainment) signed separate AI music licensing deals with Los Angeles-based startup, Klay Vision on 20 November, 2025. Recall the lawsuits against Suno and Udio mentioned earlier? Well, turns out that those are quietly being resolved through renegotiation. Corporations clearly have no issue with generative music AI itself, just with inappropriate compensation. 

In conversation with Culture Custodian on ethical ways to use music AI, The Beatoven, who composed for Cowbellpedia and The Next Afrobeats Star talent show, concurred that AI is just a tool, with permission and compensation being the major concerns. He also believes that talent still makes the difference.“There are situations where I have an idea but I can’t execute it immediately because of physical or logistical constraints. For example, you may need a choir but can’t get one into the studio quickly enough. In cases like this, AI can help you generate a rough version of your idea, which you can later refine with real musicians or collaborators. ( Take Charlie Puth’s Changes for instance. The choir is entirely A.I, he voiced it and ran it through Suno),” he explained via X.

He added, “I think it’s fine to use AI for general tasks, idea generation, or speeding up parts of the creative process. But when AI models are trained on real artists’ work without their consent or compensation, that’s where I have an issue. For instance, if someone asks an AI to make them sound like Drake or Taylor Swift, and the AI produces something that closely mimics their voice, style, or artistic essence, then we have to ask: where did that ability come from? It came from years of those artists creating music and building a unique identity. If their likeness is being used to create new works without their permission, and other people profit from it while they get nothing in return, then I don’t think that’s ethical.”

The Beatoven’s recommendations are straightforward: education and stronger copyright regulations first, and then investments in music technology startups not far off along the line. Nigeria’s entertainment law ecosystem, while not in its infancy, is still far off from the global standards that would enable answers to questions of ownership and compensation. Additionally, tools and businesses encompassing film scoring, advertising, jingles, brand identity, gaming, sound design, and many other areas where technology and creativity intersect are vital. He draws a parallel between Nigerians who listen to foreign music in other languages and outsiders who connect with our indigenous genres, showing that the human connection plays a key role, hence the need to preserve cultural heritage. “As AI becomes more advanced, in the future, it may become harder to tell what is human-made and what is AI-generated. That means our identity, our stories, and our unique sound become even more valuable,” he emphasises.  

Producer David ‘ePianoh’ Tope captures this sentiment of talent as a divider best. Known for his instructional videos on YouTube—where he has an astounding 3 million plus subscribers—and a mix of self-insert reposts, educational music production content, and hyper-animated comedic takes on Instagram, Facebook, and elsewhere, the Nigerian producer has carved a niche for himself in the generative AI music education space. His “Swap Suno Vocals For Your Own”, “How I Turn My Lyrics Into Choirs Using AI,” and “How To Make A Full Choir With Just One Voice/Controlla AI” tutorials make a case for initial talent as the core of production. 

In the video where he creates a choir using the Controlla AI tool, ePianoh’s delivery is still the deciding factor in how the finished product sounds. This is even as some of the vocal presets—661 in total, surprisingly—are audibly trained on real human beings (“Pixie” sounds like a cross between Dua Lipa and a country singer that’s just out of this writer’s immediate recall). Unnerving as the sequence was, listening to the choir sing, his lack of emotion translated through, accent and pitch changes notwithstanding. 

In another video where he transforms an entire reference track into one he sings, the finished product differs  significantly from the initial record. Imitated riffs and melodies still do not make it the same song. Perhaps a more accomplished singer might work wonders with Suno, stripping away imperfections and taking only the bits that work best. But that’s exactly what regular music production software like Logic, Fruity Loops, Ableton, Cubase, and Pro Tools already do. Which again proves that the core concern is consent and compensation. 

All the art that has been stolen to create these tools, and all the artists whose decades of ingenuity have been distilled to help yet another 12-year-old marshall a hip-hop version of some abominable nursery rhymes, has to be accounted for. ePianoh might be able to write and arrange music a lot better as a result of generative music AI. Channelling Hugh Masakela’s horns or Fatai Rolling Dollar’s sectioning might become five-fold easier too. This is a possibility. Their respective estates won’t be compensated, though. And that’s a conflict of interest we can’t build our way out of. 

When is AI music not large language mediocrity? Music is a sum of so many parts: kinetic stories, experiences, and ideologies wrapped in one. But sentiment alone isn’t enough to drive a powerful record. In a section of one of this writer’s favourite pieces of music criticism ever, writer Jeremy D. Larson harangues the emptiness of singer Benson Boone’s debut album, American Heart, by saying, “Instead of a musical or narrative point of view, Boone relies on speaking his truth, a songwriting axiom that doesn’t take into account whether someone’s truth is fundamentally boring or has been rendered in pop music countless times before.” 

And this is a similar itch with the belief that AI music is justified since it evens out access. Too many people have ‘truth’ to sing that the world has heard a thousand times before. Truth that’s as boring as NTA’s so-called documentaries. They think because it’s their truth, the world should hear it, without bothering to do the work required to make their truth unique. Not even AI can save the lack of soul. Not even AI can take away the needlessness of their soulful wailings. The ethical way out for such people is that their art never sees the light of day until it has been refined—an exaggeration, but only to emphasise that art being human isn’t enough.