<aside> 📍
You are not lost.
You have arrived at a sub-page in the AIxDESIGN Archive.
This is a public repository - our way of working in the open and sharing as we go.
Have fun!
</aside>
<aside>
INDEX
</aside>
A recap of our workshop with Mateusz Mieciek (Jak Woda) at NDSM, Amsterdam.
<aside> 🪗
Open Culture Tech is an initiative by Thunderboom Records and the Netherlands Institute for Sound & Vision that makes immersive technology more accessible to artists through residencies, open-source tools, and educational showcases. As part of OCT 2.0, AIxDESIGN partnered to host a series of workshops exploring how AI shows up in music production - both its creative potential and the trickier questions around agency, authorship, and how artists actually want to work with these tools.
</aside>
In May 2026, we hosted media technologist and artist Mateusz Mieciek (aka Jak Woda) for an in-person session at the AIxDESIGN Foundation, NDSM, Amsterdam. Mateusz had travelled from Copenhagen to lead the workshop, and the turnout was on the smaller side - which meant the whole afternoon could move at conversational pace. Quality time over crowd. You can read the full workshop description here.
Mateusz works with phygital tools, embodied musical interfaces, and AI noise augmentation (the opposite of noise cancellation, which is a very Jak Woda kind of inversion). He's been producing and releasing music since 2019 and runs workshops at festivals including Roskilde. The thing he keeps coming back to: tools should invite creativity out, not gatekeep it.
The pitch he'd written up in his application stuck with us:
"We don't want to learn how to make music farms (which is actually pretty simple). We want to become cyborgs and create merging with technology." - Mateusz Mieciek
That distinction - farms vs. cyborgs, output vs. entanglement - shaped the whole afternoon.
The workshop moved through a deliberately janky multi-stage pipeline: body to sound, sound to sound, text to MIDI. Each round used a different kind of intelligence (human gesture, audio model, language model) and each round broke in interesting ways.
Round 1: Gesture → Audio. Mateusz started with his own browser-based p5.js sketch - an "embodied player" that uses webcam-based hand tracking to trigger PolySynth sounds. Open palm, pointed finger, distance to camera: each gesture mapped to a parameter. As Mateusz pointed out during the session, hand recognition is actually the most basic thing AI does. It recognises a pattern. That's it. But used as a musical interface, it becomes a digital theremin you don't have to buy or build.
Round 2: Audio → Audio. Participants took the gesture-sounds they'd made and fed them into EzAudio on Hugging Face, prompting in plain text to transform the source material. The "alien" framing Mateusz had set up at the start took over almost everyone's prompts. "It's hard to get out of the alien vibe," one participant noted, somewhere between observation and admission.
Round 3: Text → MIDI. This was the strangest leg of the journey. To get a language model to generate music, you can't just ask for "a beat." You ask it for ABC notation - an old plain-text music format from the 1970s that LLMs can actually produce. Then you convert the text to MIDI. The model generates drums, bass, and melody as separate text blocks, often forgetting line breaks, sometimes producing something playable, sometimes producing what sounded like a computer pacing the room.
By the end, everyone had a folder of fragments: gesture recordings, EzAudio textures, MIDI snippets. The plan was always to build something out of those scraps together.
The angle that emerged most clearly came in the closing reflections, when one participant - more visually than musically trained - named what she'd been doing all afternoon. The structure of the workshop, she said, mapped exactly onto her practice of collage: gather scraps from existing material, treat them as a dataset, recombine. The unfamiliarity of working with sound dissolved once she could see it as the same operation, in a different medium.