Humans in the Loop

Remember the 70s? Flared trousers, questionable perms, and the revolutionary arrival of the synthesiser. Suddenly, electronic squiggles and otherworldly washes of sound were creeping into everything from prog rock epics to disco bangers. Purists probably scoffed, muttering about the death of “real” music, but the synth, for all its bleeps and bloops, was ultimately embraced. Queen proclaimed loudly in every album up until the game that no synths were used. It was a tool, an extension of human creativity, adding new colours to the artistic palette. Does that make the AI bands today any different?

I think not. We’re not just talking about a fancy new keyboard anymore, are we? We’re talking about algorithms that can churn out tunes with alarming proficiency, mimic voices with eerie accuracy, and even conjure up entire “bands” complete with faux-biographies and pixelated press shots. Enter The Velvet Sundown, the latest poster children for this brave new world, a band so convincingly AI-generated they managed to fool a fair few before their synthetic origins were revealed.

I am a musician. A very bad one at that. I use keyboards. I use GarageBand. I use samples. Before you accuse me of being a Luddite clinging to my dusty vinyl, let’s be clear: using AI as a tool is perfectly acceptable. In fact, it’s exciting. Imagine a composer struggling with writer’s block, using AI to generate a hundred variations on a theme, picking the best, and then twisting it into something truly profound. Or an artist, perhaps one whose voice has been ravaged by illness, using AI to rekindle their vocal cords, giving them a chance to perform again (The Beatles’ recent “Now and Then” is a perfect, poignant example of AI as an enabler, not a replacement). ABBA Voyage is an absolute delight. It is a visual and sonic masterpiece.

This is AI in its rightful place: augmenting human talent, extending human capability, providing a kick up the backside when inspiration falters. This is where Humans in the Loop are paramount.

Some would argue, for all its technicality, music’s greatness ultimately lies in human connection. It’s the raw emotion poured into a lyric, or the shared experience of a band honing their craft in a dingy rehearsal room. These are the narratives that resonate with us, the stories that make a song more than just a sequence of notes. Where’s the documentary on the making of [insert AI produced claptrap here]. When we connect to a piece of music, we somehow understand it comes from a place of human experience – joy, sorrow, anger, love. It’s a reflection of our humanity. Isn’t it?

The problem arises when the shiny, perfectly sculpted track dropping into your Spotify feed isn’t from a group of earnest musicians, but a cunning algorithm disguised as one. When an “artist” has no physical presence, no tour dates, no awkward interviews, just an endless stream of suspiciously consistent output and AI-generated album art that screams “stock photo with filters,” a cynical alarm bell starts ringing. It’s the musical equivalent of being catfished, and frankly, it feels a bit, wrong.

The Velvet Sundown released two albums in jig time. That’s a flag right away. And their creator even tried to blag off the initial accusations. Cynically some saw it as Spotify looking to get even cheaper musical rights. If you have seen their artist reward model you’d not be far off thinking that was already impossible!

For me, this isn’t just about protecting the sanctity of the human part of the creative process; it’s about transparency. When a product is marketed as one thing but is, in fact, entirely another, consumers have a right to know. If I buy a delicious-looking apple pie, I expect it to contain apples, not some synthetic fruit-flavoured goo.

So, here’s my modest, slightly jaundiced and biased proposal: let AI continue to be an incredible tool. Let artists and producers experiment, push boundaries, and discover new musical frontiers and landscapes with the help of intelligent algorithms. But when the music is entirely, or even predominantly, the brainchild of a non-sentient entity, then let’s be upfront about it. Streaming platforms, take note. A simple, unobtrusive flag – a tiny “AI-Generated” tag next to the track title, perhaps? – would suffice. It wouldn’t diminish the technical achievement, but it would honour the truth. Informed consent and all that. Have a setting that says “no AI-Generated” on the service as well. Imagine if you had a “no synth” setting. You’d have missed out on Another One Bites the Dust.

I digress.

At the end of the day, while AI can learn to mimic, to synthesise, to mimic and create adjacent synthetic music, it cannot yet genuinely feel. Yet.

Until it can weep over a lost love, rage against injustice, or create the shared experience of four schoolkids going from first rehearsal to first gig to first hit, then the most profound music will arguably always be found where Humans are firmly in the Loop. Let the machines play, but let’s label them as such.