Museums Night ’25 with My Name is Fuzzy

Isabella: Who is Fuzzy? How did the persona come to life?
Fuzzy: Fuzzy is like a version of myself that just became a character. It started about five years ago when I was making music installations and needed someone to feature in them. I wasn’t sure about using myself, but then I decided to do it. In different projects, Fuzzy became a character, like an old music star from the past. It’s funny because for me, it’s really me—just me playing different roles. But for other people, it became a character, and I play with that. I don’t really make the difference between us, it’s just that Fuzzy is on stage and I’m not. Fuzzy is a good character I can use to tell any story because it can be both personal and fictional. So, I like it.

I: And the name?
F: Fuzzy is a very old nickname from my first band (The Rambling Wheels) in the early 2000s. I was a bad guitar player, so I used a fuzz pedal to hide it. I also had a cap that said “My name is Fuzzy.” The day I started my solo project, I had no idea what to call it, but I saw the cap and thought, “Okay, so my project will be My name is Fuzzy.” I like it because “fuzzy” also means a bit blurry, not very clear. My projects sometimes blur the lines between different things, so it fits.

I: During Museums Night 2025, we showcased your work La machine à tubes [The Hit Machine]. Can you tell us about it?
F: I had the idea before AI was so widely accessible. My previous projects were always about finding new ways to show pop music. I was making pop music, but I didn’t want to keep on touring, recording singles, and following the usual industry patterns. I wanted to be more free in how I could show my music, and that’s how I started doing music installations. First, I had the idea to make a digital double of myself, a way for people to write songs in my place. Initially, it was just about having samples that people could launch, creating new songs each time. Then, when I started digging into the project, AI was becoming more accessible. So, it turned into creating a double of me that could write songs for people live.

It’s about the image of an artist always chasing the fantasy of writing a hit. This machine does the same—trying and, trying, but never knowing if it will actually write a hit. That was part of the concept. I really liked early AI because the imperfections made me laugh. When a “robot” does something wrong, it triggers something between amusement and pity. The broken side of it says something about the topic and also about how fast technology evolves. What’s funny is that the older the installation becomes, the more successful it is. I thought I’d show it for six months, and then people would lose interest, but it keeps drawing attention.

«La machine à tubes» 2023, My Name is Fuzzy during Museumsnacht 2025 at HEK. Photo: Moritz Schermbach

«La machine à tubes» 2023, My Name is Fuzzy during Museumsnacht 2025 at HEK. Photo: Moritz Schermbach

I: How does it work?
F: All the lyrics are based on things I’ve said in interviews. When we started the project, training an AI on my lyrics wasn’t possible because I hadn’t written thousands of songs. Feeding it just 40 songs didn’t give usable results. So instead, we took interview transcripts, picked sentences, and used them as topics for the AI to generate lyrics.

The melodies are inspired by my existing songs. If someone selects “sad” on the terminal, the system takes a melody from one of my sad songs and runs it through an AI to create something new. There’s also some randomness—like using sampled sounds from the keyboard I play the most. The goal was to make it sound like demos I could have made.

For lyrics, we used ChatGPT with an API. For melody, we used Magenta, an old AI from Google, almost 10 years old now. When the project started, AI couldn’t generate singing voices, so I made a voice clone in a special way—I recorded myself speaking in a monotone for an hour, then ran it through live autotune to make it “sing.” The whole thing was built with accessible, cheap tools since we don’t have the resources that a big studio does.

I:Do you view AI models as collaborators, tools, or something in between?
F: A bit of both. I use AI often for small things—either for practical reasons or because I enjoy the creative side. But I also see its limits. Right now, I don’t think AI can write interesting lyrics—it’s not creative enough. But for factual things, like rewriting applications for different forms, it’s super useful. Even for melodies, AI feels like having another musician in the studio. I’m not afraid of being replaced by AI in the art world. It’s more of a tool or a collaborator we can work with. What’s interesting is what we do with it, not what it does on its own.

I: Nostalgia seems to pop up in your work through your aesthetic language. You’ve described it previously as retro-futurist. What significance does this have to you?
F: I think I just have “old” tastes. Every time I design something; I naturally lean toward retro aesthetics. I’m not nostalgic—I don’t long for the past—but I like mixing existing styles with something new. That might make me feel a bit “off” in the contemporary art scene, where everyone wants something never seen before. But I like working with things that are familiar, things people can relate to. My work often has an ironic distance, and I think that applies to my design choices too—I observe things from afar, including time itself.

I: Is there a piece of media – whether YouTube video, artwork, music, meme, etc. –, that you think needs to be seen by the world?
F: One comes to mind—Soundtracks at the SFMOMA in 2017. It was the first time I saw music being exhibited, and it really gave me confidence in approaching music this way. One work that particularly struck me was Visitors by Ragnar Kjartansson. It featured large screens in different rooms of a house, with musicians playing together. That exhibition, along with discovering Bill Viola’s work, showed me that I could continue doing pop music but stage it differently. It pushed me in this direction.

I: What’s next for Fuzzy—or for Bastien? Any exciting projects on the horizon?
F: With the french music producer Corentin Kerdraon, we are selecting 10 songs the Machine has produced (among 4’000) and are re-recording them and rearranging them with musicians, in order to make them become “real” songs. The idea is to close the loop of the machine inspired by me, then creating songs generated by the audience and finally reinterpreted by humans. That will be an album I haven’t written, but provoked.

Next year, I’m working on an AI-driven improvisation show with two actors. The audience and AI will create a PowerPoint on an absurd topic, and the actors—who haven’t seen it—will have to present it live. It’s a mix of theater and chaos. I’m also diving into electronic music, working on a solo live techno set—no computers, just machines, fully live and somewhat improvised. It’s very different from my usual staged performances.