This series may be called 'The Best Music of 2025, So Far', but we have to briefly take aim at a not-so-great emerging trend from this year too. See, if you take credulous LinkedIn goobers at their word, generative AI is Radically Changing Everything™. Soon, we’ll be living in an incredible ‘agentic’ future, where Siri will book vacations for you while Gemini takes your kid to daycare and GPT-5 once released will complete the construction of the Sagrada Familia. (Don’t go off on me about how large language models work — I know I’m being hyperbolic here, but tell me this is actually that far removed from some of the shit these people claim). Of course, none of this is actually happening anytime soon, but you know, it could any minute now! — so please give all the companies behind them money, lest you miss out on the big new thing, just like you totally missed out on the metaverse or NFTs or whatever.
As part of the race to milk every craven opportunity this technology allows, generative audio startups like Suno and Udio also now have come into existence, borne out of the timeless question: "How can we burn venture capital funding to auto-generate soulless slop by sampling from a dataset of stolen existing music?" And so anyone can now write a simple descriptive sentence about a style of song and have a resource-guzzling data centre spit out a cheap imitation of music—output that may even be believable at first listen, but is bereft of any real human touch, save for whatever the model can regurgitate from the stolen material it was trained on. (And no, prompting a model does not make you a musician, the same way me paying a session saxophonist doesn’t make me a saxophonist; and yes, this is the kind of thing that should be gatekept.)
All this might have just been a frivolous novelty not too long ago, but we’re now at the stage where wholly-AI-generated music is a real thing that people are being tricked by. Generative models for music have existed for a while now (remember the neural network trained on Archspire music that ran endlessly on YouTube?) but advances in compute coupled with tech bros’ willingness to just steal content have led to increasingly realistic model output. This means that some charlatans are all too happy to farm these models for slop, release it on streaming services, and lie about what created the music, of course, because they know that the whole concept is fraudulent and off-putting. A "band" called The Velvet Sundown was in the spotlight recently after word came to light that their existence was entirely AI-generated, while the human-ness of black metal project Draugveil’s album continues to be the subject of debate. In an especially vile turn of events, you can even find AI-generated fake music uploaded under dead artists' names on Spotify, trying to mimic the work they put out while still alive. How soon before some dipshit uploads AI-generated garbage to Spotify and claims they're unreleased Ozzy tracks? And for what? Pennies in streaming revenue?
We could get into a bunch of topics here: the environmental impacts of running these big generative models, what they mean for copyright and ownership, what they say about musical expression more broadly, or how they cheapen the various meaningful purposes music plays in our lives. Instead, I’m just going to keep it simple: this shit sucks. Everything about this stinks, and at this point, anyone still trying to pass off AI-generated music and artwork as genuine should be mocked (although only up to a point, I’d argue, because even an internet pile-on is still a form of attention). If you can’t be bothered to make it yourself, no one should bother listening to it, let alone engage with it as if it's a genuine act of creative expression. Don’t let this garbage be normalised, as much as its acolytes love to insist it’s already a done deal and that there’s no going back. Resigning oneself to that is how we have musicians like Steven Wilson catastrophising in Facebook posts about an inevitable future where musicians are not needed and people will generate "personalized music" for themselves. It’s repulsive to imagine, but also an entirely avoidable outcome.
There’s probably going to be more slop being made for now, which gives us all even more reason to instead celebrate the tidal waves of incredible music being put out by actual artists. We have hours of listening material listed through this mid-year review series, and even more in the weekly Release Day Roundups. So, I would recommend we continue depriving gen-AI trash of the oxygen its talentless, lazy creators desperately want for it, and continue uplifting actual music however we best can. Above all, do not fall for the posturing about how this is just the new normal that we need to adapt to. It really, really doesn’t have to be.