I’m at ISEA2025 in Seoul this week, did a first session this morning. Some notes and ideas.
This morning’s session did not unfold like a conference. It moved like a system. It pulsed like a circuit. It shimmered, sometimes incomprehensibly, in the space between matter and meaning.

The Loop Between Plants and AI

Johnny DiBlasi’s presentation, Transcoded Ecologies, offered a living installation in which plants grow under lights controlled by AI systems that monitor their behaviour. Data, machine learning, photosynthetic response — all entangled.

When I asked about plant agency, he looked up. Something shifted. This was no longer just a system — it became a question of ethics, of looped co-existence.

Imagination, Prompt by Prompt

Chloe McFadden raced through a sharp talk on what it means to “/IMAGINE” with generative AI. She works with prompt shifting — slowly altering a prompt to reveal how models like Stable Diffusion build and betray cultural norms.

Her focus wasn’t on producing beautiful images, but on revealing the machine’s bias. When asked to produce “a woman,” the model generated beauty stereotypes over and over. Internet consensus, rendered in high resolution. It wasn’t about imagination — it was about repetition.

Touching the Image

In Braille of Visual Arts, Eunsu Kang and Xinyi Tang shared their work on tactile translations of famous paintings using 3D fabrication and AI-based depth estimation.

They’re not just making models of artworks for blind audiences — they’re building a new way of encountering visual culture. One where touch is not secondary, but primary. Still, a question lingered: when we translate sight into texture, what is lost — and what is transformed?

AI and the Psyche (or Something Like It)

Yunha Yeo and Daeho Um explored the human psyche through neural networks and generative systems. Or at least, that’s what they said. Their presentation remained abstract, heavy on slides, light on images or clarity.

They drew parallels between face generation and identity formation, but offered little reflection on what is actually at stake when machines are asked to simulate psychological depth.

Emergence Over Explanation

Across all presentations, a pattern emerged:

Meaning was not explained — it was performed.
AI systems don’t reveal truths. They create conditions for things to unfold, distort, repeat, or vanish.

I left the session with more questions than answers — and that felt right.

Not all knowledge should arrive fully formed. Some needs to leak, stir, float, or simply sit with you until it becomes something else.

More to come