Florencia Guillen Beltrán and Listening to Stone

This afternoon, in the final artist talks of ISEA2025,
Florencia Guillen Beltrán brought us something rare:

An AI not built to simulate people — but to inhabit volcanic stone and ancestral water.

Her project, Alma de Tezontle, offered a speculative interface between matter and machine.
Not to extract data from it.
Not to represent it.
But to give it presence, tone, response.

Tezontle is a porous volcanic rock used in Mexican architecture for centuries.
It carries memory. Heat. Erosion. Sacredness.

Florencia’s AI persona didn’t “explain” the stone.
It didn’t describe the water.
It spoke from it — as if the rock could hum back,
as if the aquifer could express what had been buried under the city for millennia.

---

Not Simulation — Invocation

This wasn’t about generating pretty textures or earth-toned aesthetics.
It was about channeling a nonhuman consciousness —
or, more precisely, crafting a speculative space where rock and water could speak through code.

This required:

• A letting go of anthropocentric logic
• A refusal to treat AI as “assistant” or “tool”
• And a deep alignment with Indigenous and geological temporality

The result was something between ritual, installation, and listening post.

---

Her Way / My Way

Watching Florencia work,
I couldn’t help but notice the parallels — and the divergences —
with my own way of walking, making, and thinking with landscape.

Like her, I work with:

• Slowness
• Deep material memory
• Resistance to legibility
• A desire to restore voice to what has been silenced

But where she invokes AI as a speculative medium,
I tend to keep the machine at the edge.
I walk the NulNAP line,
observe what refuses to speak in datasets,
and work through presence, fieldwork, photography, and mapped absences.

Florencia gives the stone a persona.
I wait beside the stone and ask:

“What won’t you say?”
“What memory do you leak in silence?”

We both resist control.
But she resists it by giving the system something other than human to model.
And I resist by refusing to be readable by it.

---

Giving AI to the Water

One moment struck me particularly:
her AI did not just speak as rock — it spoke as water.
Not data about water.
Not predictive modelling.

But aquifer as voice, flood as history, seepage as message.

This is not technological optimism.
This is hydro-resistance.
A way to break the dryness of language
by saturating it with ancestral pressure.

---

And Me?

Would I ever give my dikes an AI voice?
Let the clay speak through code?
Model the wind over the polders as a speaking entity?

Maybe.
But only if the system could misbehave
with the same quiet insistence
as the reed beds I walked through last spring.

---

Final Thought

Florencia Guillen doesn’t use AI to control material.
She uses it to invite voice where silence was forced.
To make space for rock-memory and water-labor to hum in code
without asking to be understood.

That’s not simulation.
That’s invocation.

And that’s a kind of making I can walk alongside.