Hiya Yulmi, thanks for joining us in the virtual RGM lounge today. Grab a brew and take a seat.
Thanks for joining us today. Where are you from, where do you live now, and how did Garbage Garden begin?
I’m originally from South Korea, and I currently live in Kobe, Japan.
I started working as Garbage Garden around November 2025. The trigger was communication with a young user I got to know through a community I belong to.
Making a song for that user was kind of an awakening for me, I guess.
Are there any artists you think we should be looking out for?
Yes, of course! Wendy Lucas (IG: wendylucasmusic) from the US, Leona Rue (IG: leonaruemusic), Austrian indie band Valone (IG: valone_official), Dizzy Michal (IG: dizzymichal), UK indie folk singer Laura Williams (IG: laura.williams_music), and from Japan, HOME (IG: home_band_), Rakia (IG: rakia2002), Blackur0 (IG: kuro.musiccreator), and Ozwald Alexander (IG: ozwald_alexander).
They’re all human-led indie artists I really respect.
I’d also mention Cosmos Boys (IG: m_cosmosboys), an AI music project mainly working in math rock and emo.
What’s your favourite social media platform to use and why?
I enjoy using Instagram.
I had never really used Instagram before, but I think it’s one of the best places to discover music from lesser-known indie artists. At least, that has been my experience.
What’s currently in your ears and on rotation at the moment?
I often listen to the indie artists I mentioned above.
Phoebe Ryan has also come back recently, so I’ve been listening to her new songs a lot too.
Garbage Garden moves across a lot of different genres and sounds. Do you think about genre first, or do you choose the sound depending on what you want to express?
I usually think about the story I want to express first, then I write the lyrics, and after that I think about what kind of sound or genre would express it best.
Of course, I think I’m able to work that way partly because all of my music is AI-assisted.
You mentioned that all of your music is AI-assisted, so let’s address the obvious question directly. What are your thoughts on AI-generated music?
Since I’m using it myself, I don’t think I can simply say AI-generated music is bad.
I know some people use it in negative ways, but as a tool for expression, I think it works well.
Of course, there are also ethical issues that have come to light recently, and I do feel bad about that.
When you say it works as a tool for expression, what does AI allow you to do that you couldn’t before?
First of all, I think of myself more as someone who wants to express things than as a musician.
I actually feel a little uncomfortable calling myself a “musician,” because it feels disrespectful to people who make music without AI assistance.
Anyway, it’s not necessarily that AI lets me do something I could never do before. I think the biggest advantage for me is that it can reduce the time and cost that would otherwise be needed to turn an idea into something real.
Even saying that makes me a little cautious, though.
You said you think of yourself more as someone who wants to express things than as a musician. So what does “expression” mean to you?
For me, expression is simply giving form to something inside me.
It could be a feeling, a thought, or a story. Sometimes it becomes lyrics, sometimes an image, sometimes music.
I don’t really care about the format as long as it says what I wanted to say.
Let’s move on to your new single, “Fake,” which is coming out on August 28. What can you tell us about the song?
As AI music companies have emerged and the landscape has changed, the backlash against AI music seems to have grown stronger.
Some people even argue that it shouldn’t be called “music” at all.
“Fake” was conceived in that atmosphere, but it isn’t some grand manifesto. I’m not trying to tell people that their negative feelings toward AI music are wrong either.
To talk about myself for a moment, I’ve worked as a translator for many years. Before that, I used to make video games.
I always had the desire to express something of my own, but I became much more used to delivering other people’s words instead.
At the same time, translation is one of the professions at serious risk of being replaced by AI.
So when the same AI that could tighten its hands around my throat ironically became a tool for giving my own suppressed “words” a way into the world, how was I supposed to react to that?
That contradiction is what became “Fake.”
You said you don’t think the backlash against AI music is simply wrong. What parts of that criticism can you understand?
I don’t think I’m in a position to judge whether the criticism is fair or unfair. But I do understand some of it.
I used to be just an ordinary listener, so I honestly couldn’t even imagine what many indie artists were going through.
I only started to understand a little after I began doing self-promotion and trying to get people to actually hear my own songs.
Most of them just want someone to listen to what they have to say, but the reality is already very harsh.
And I do think the excessive circulation of AI-generated music has, to some extent, made that situation even harder.
There are also ethical concerns around AI companies that have become more widely known recently.
I’m not sure I can simply say there’s nothing wrong with that either.
So where do you personally draw the line between AI as a tool for expression and AI being used in ways that make the situation worse for other artists?
This is only my personal view, of course.
I don’t make music in the usual human-led way, but at least I try to put care into each song I make.
There was a time when I generated several songs a day too, but now it can take me months to finish just one song I actually want to release.
I’m not trying to say that this is the same level of effort as making music in a human-led way.
I just try not to let the value of each song get closer and closer to zero.
If I started uploading huge numbers of songs like spam — which I honestly don’t have the intention or ability to do anyway — I think that could reduce not only the value of my own songs, but the value of everyone else’s music too.
Do you think the real problem, then, is less about whether AI is used at all, and more about how people choose to use it?
It’s not an easy question for me to answer either, but yes, I think that could be the case.
AI is already being introduced into tools that make mastering and mixing easier, as well as web-based DAWs and other parts of music production.
So I don’t think we can rule out the possibility that, someday, AI will be involved in many songs in one form or another, whether people are aware of it or not.
Because of that, I think how we use AI, and what kind of ethical standards we have when using it, may become more important than AI itself.
You mentioned that it can now take you months to finish a song you actually want to release. What was that process like with “Fake”?
I spend quite a lot of time on Instagram and Reddit, so I was seeing more and more discussion around AI music.
Around the middle of April this year, it felt like the hostility became noticeably stronger. I also received some negative comments myself.
I hardly post on Reddit because of that, actually. (laughs)
Those experiences and observations were what made me start developing the lyrics and concept for “Fake.”
I spent most of May working on the song, and I think I scrapped the overall direction two or three times because I had chosen the wrong genre or sound at first.
Once I get a melody or structure that feels close to the image I have in mind, I keep working from there — doing some DAW editing, adjusting the instructions for different lyric sections, reviewing the overall prompts, and generating again.
It’s a repeated process of listening, adjusting and rebuilding.
By early June, I finally had a version I personally felt satisfied with.
You said you scrapped the direction a few times before “Fake” finally felt right. What kind of sound were you trying to find for the song?
At first, I imagined something closer to an early track of mine called “my_fragments.exe” — avant-pop mixing pop rock with 2000s nu metal.
I even hummed melodies over the first lyric draft and thought it might work.
But it ended up feeling too flat. The arrangement was a little chaotic, and the melody made the narrator sound more sarcastic than I wanted.
Then I tried going in the opposite direction and making it brighter, but that didn’t work either.
I also realized the lyric structure itself was part of the problem — originally, the main chorus didn’t come in as early as it does now.
So I changed the structure and moved toward metalcore / post-hardcore with some emo influence, using vocals that felt harsh but emotionally fragile.
Looking back, I don’t think I was searching for one specific sound from the beginning.
I found the sound while making the song.
You described the vocals as “harsh but emotionally fragile.” Was that contrast important to the character or narrator of “Fake”?
Yes.
As I said earlier, “Fake” isn’t a manifesto, and it isn’t a song that denies the feelings or opinions of people who dislike AI music.
It’s more like:
“Yes, maybe everything you’re saying is right. But this has already become my ‘mouth.’ What am I supposed to do? Am I really a criminal just because I want to express myself?”
I needed that feeling of someone almost screaming those words from inside the song.
When you’re not working on music, what do you usually do to switch off?
I usually cool down by doing my day job — translation. (laughs)
What’s next for Garbage Garden after “Fake”? Anything we should look out for?
Actually, after “Before the Magic Fades,” I thought I would stop telling my own story for a while.
Garbage Garden originally started from more of an “observer” position anyway.
Since I ended up returning to my own story again with “Fake,” I think I’ll go back to that observer position after this and continue exploring relationships and existence.
Thanks for joining us today, Yulmi.
IT’S FESTIVAL SEASON, ALL THE LATEST LINE UPS AND FESTIVAL NEWS HERE AND INTERVIEW NEWS HERE