There is a resistance to memory inside the memory itself. 
A Symphony in Five Acts (2024)


MAY–2024
Incubation and presentation of first two acts at FEY ARTS FESTIVAL, Chateau du Fey
Group Show 

1 - 5 OCTOBER 2024 
CIVA, coded manoeuvres
+ EXHIBITION 
01 OCT 2024 - 26 JAN 2025
BELVEDERE 21, VIENNA





There is a resistance to memory inside the memory itself.
– Jacqueline Rose 

+ + +

ACT ONE 
{WE ARE OBSERVING EACH OTHER}

1/ [AI generated video, 11:40 min, loop, voice & soundscape]

2/ [Face recognition / Camera capture / AI visitor-character live generation in co-creation with self-trained models and alias LoRA]

3/ [LLM live questions generation]

4/ Imprint of the visitors in the latent space

ACT TWO 
{SANCTUARY OF CO-INDIVIDUATIONS}


[AI-generated mantra, video work, 30:00 min, loop, poetry, voice + soundscape]

+ + +

A breath ~ a breath ~
branching paths myceliate,
skin shimmers with
each breath becoming
you. A breath ~ 


exerpt from Erodium* by Elisabeth Sweet 

Gaps in memory prevent acceptance of the past and deny connection to the future. Forgotten and repressed memories amputate lived experiences and entire histories, thus concealing the interdependence which co-evolved among species. Gaps in language further relegate groups to patterns which normalize separation and imposed hierarchies.

In these gaps, these hidden spaces, the most indeterminate, undefined aspects of existence find sanctuary. These sanctuaries are protected by forgotten and semantic paths which connect species beyond language. Sanctuary is inaccessible because it is forgotten; it is beyond reason because it exists in the non-linguistic territory of a shared consciousness.

Machine learning models allow humans to explore the latent space of language through non-linguistic means. Visually evolving instantiations override static notions of what is and reveal the possibilities of what could be, what might have been, and what we may see if we shed reliance on human language as a method of connection and understanding and instead consider different forms of intelligence as incubators for speculative futures.

Drawing on the fundamentals of Gilbert Simondon’s philosophy, which emphasizes the genesis of individual entities through processes of individuation rather than predefined categories, the project envisions sanctuaries as spaces of becoming, encouraging a varied forms of radical empathy. 

+ + + 

CREATION METHODOLOGY
The visual elements within the films were synthesized using open source models alongside our custom model fine-tunings - trained on the essence of our physical collage work - and bespoke technical pipelines. These resultant visuals formed the foundational dataset employed for training our alias*. 

The films themselves emerge as manifestations of conscious exploration through the latent dimensions of said model.

The poem, *Erodium, written by Elisabeth Sweet uses the breath and tempo as a method of access to memories of assault. By channeling pain into rhythm, the memory dissolves into memory, shedding resistance and revealing light. 

The mantra-like soundscape is composed by a musician and violinist Sayaka Botanic

- - -

*More about alias: an alias is a model trained and approved by the artist, based exclusively on licensed data sets. The alias studio founded by a team of AI scientists, entrepreneurs, legal experts and researchers, enables the creators to turn artistic expression into unique, certified generative AI models.

The studio invited us to be among the first artists and ambassadors to experiment with the platform. 


 1/ Still from We Are Observing Each Other, AI-Video, 2024


2/ Face recognition / Camera capture / AI visitor-character live generation in co-creation with self-trained models and alias LoRA, 2024


4/ Imprint of the visitors in the latent space, AI generated video, 2024


Still from The Sanctuary of Co-Individuations, AI-Video, non-verbal travel through a CROSSLUCID’s self-trained model, 2024

Excerpt from The Sanctuary of Co_individuation, AI-Video, non-verbal travel through a CROSSLUCID’s self-trained model, sound & voice, 2024