Interactive Artwork Generative Poetry · Browser-Native · 2026
BORROWED
SIGNALS
Text you don’t read — you operate.
An interactive poetry-reading experience where the text becomes something you operate: a gallery of generative poems you scan, tilt, speak, and hold to read. The work is split across your phone and your desktop on purpose, so you never see all of it at once — a piece about surveillance and borrowed language, built entirely in plain web technology.
Reading, made into something you do.
Borrowed Signals is a gallery of interactive poems — short works, and not one of them sits still on the page. You scan, tilt, speak, drift, and hold to read them; each poem only gives itself up in exchange for your attention and your movement.
At its core it is an experiment in engagement. What happens to a text when reading becomes something you do with your hands, your phone, your whole body, instead of something your eyes merely pass over? Every piece is a different answer — a different way to experience language when the page can watch you back.
One wall. The same question, asked from every angle.
Every piece answers one question: what do you do when you’re scanned, displaced, and optimized? The work lives where street art meets surveillance — concrete and spray paint on one side, scan-lines and tracking reticles on the other, all filed by the same cold hand as a registry of measured specimens.
The gallery is split across your devices on purpose.
The screen in your hand holds one half of the work. The screen on your desk holds the other. Whatever you are looking at, the rest is somewhere you are not — so you never see everything at once. That is not a technical limit; it is the piece. It is exactly how being watched feels from the inside: a curated slice, handed to you, with the rest withheld.
Interaction, generative language, and plain web.
You are the instrument. A cursor becomes a scanner beam; a phone’s tilt becomes gravity; your voice gates a reveal; your face, through the camera, wears the poem on its own mesh. The reading only happens when you move.
The poems are written with generative language models — and that choice is the meaning, not a shortcut. A model doesn’t confess; it predicts the most probable continuation of everything people have already written. To read generative poetry is to read a statistical portrait of us: not what a machine feels, but what we, on average, are most likely to say. Borrowed Signals stands in the gap between probable and wanted, and asks you to feel the difference.
And it is all plain web. Every work runs by opening a single file in a browser — HTML, CSS, and JavaScript, with Canvas and WebGL for the moving parts. No build step, nothing to install. The art is the source.
- ⊹Scan — cursor / touch as a surveillance beam that resolves text from the dark.
- ⊹Tilt — device orientation moves words, gravity, and ground.
- ⊹Speak — microphone energy gates the reveal (no speech is recorded).
- ⊹Face — camera maps the poem onto your own facial mesh.
- ⊹Hold & drift — dwell, steer, and harvest the text to read it.
- ⊹Split — 6 works on desktop, 6 in the pocket; one ghost you can’t reach.
A personal note.
I keep returning to a single uneasy feeling: that so much of what I see, say, and even want has already been measured, predicted, and sold back to me. We are scanned at borders and in feeds; our attention is optimized; our language is increasingly written, completed, and corrected by machines trained on everyone at once.
Borrowed Signals is my attempt to make that feeling something you can hold rather than just argue about. I wanted work that doesn’t lecture, but puts you inside the experience — a little surveilled, a little complicit, a little moved — and lets you decide what it means. If a poem written by a probability machine can still make you pause, that pause is yours. That is the one signal in here that isn’t borrowed.
Noam Sa-Man
Multimedia artist · Experience designer · AI explorer
Noam Sa-Man works at the seam between people, screens, and the systems that watch them. As a multimedia artist and experience designer, he builds interactive work that turns ideas into things you move through rather than things you’re told. As an AI explorer, he treats generative tools as collaborators and as subject matter at once — using them to make the work, and making the work about what it means to use them. Borrowed Signals is where those threads meet.
nosaman.com/art →Desktop works · 1 of 2 instruments




Pocket works · 2 of 2 instruments




Walkthrough video
Best experienced live The work is interactive and split across devices — the truest walkthrough is to open it on a phone and a desktop at once. A recorded capture can be added here for press who can’t run it.Technical notes
| Type | Interactive web artwork · generative-text gallery, 12 short works |
| Stack | Plain HTML / CSS / JavaScript. Canvas 2D + WebGL (Three.js via CDN). No build step, no framework, no install — each work is a single file opened in a browser. |
| Sensors | Camera (facial mesh), microphone (energy only, no speech recognition), device orientation / gyroscope. Permission-gated; graceful ghost-mode fallback if denied. |
| Text | Generative language models — used both as the writing method and as the subject matter (poetry as a statistical portrait of collective language). |
| Delivery | Runs from any static host or locally. Sensor pieces require HTTPS or localhost. Desktop + mobile served as deliberately split halves of one work. |
| Display needs | Two browsers (one phone, one desktop). Headphones optional. Works unattended; resets on reload. |
For editors & curators.
Borrowed Signals is a finished artwork with a thesis, not a prototype. It sits at the intersection of several beats at once — which is the point.
- Interactive & embodied: you read with your body, not just your eyes.
- Browser-native: no app, no install — open a link and you’re inside it.
- Split across devices: a structural metaphor for surveillance.
- Borrowed language: AI poetry as a mirror of collective speech.
- Made in plain web tech: the medium and the method are the same gesture.
Experience the work Open Borrowed Signals →
Artist & press nosaman.com/art →
Stills High-resolution images available on request via the site above.
Suggested credit Noam Sa-Man, Borrowed Signals (2025) — interactive web artwork.
you haven’t seen it all.