Skip to content

Art · Creative AI / Artistic Research

AI Creative Practice / Speculative Archive

An independent 2026 artistic-research strand examining how generative systems reconstruct memory without lived experience.

01

Context

An independent artistic-research strand begun in 2026, working with fragmented text, archival forms and unstable generated imagery to explore absence, inherited trauma and the limits of computational understanding.

02

Question / Challenge

What happens when a system with no lived experience is asked to reconstruct memory? Where does plausible reconstruction end and fabrication begin?

03

My role

Independent artist and researcher: concept, research, image production, sequencing, editing and written framing.

04

Process

  1. 01Assemble fragmentary source material and archival forms as the starting condition.
  2. 02Prompt generative systems for reconstructions of scenes that have no photographic record.
  3. 03Record where output becomes confidently wrong, and keep those failures as material.
  4. 04Edit and sequence images and text into archival structures rather than single pictures.

05

AI contribution

Generative systems function as both medium and object of critique — producing the speculative imagery while demonstrating, in their errors, what computational reconstruction cannot reach.

06

Human judgement

Selection, sequencing, framing and refusal. The decision about what is shown, what is withheld and what is described in words rather than images is entirely authorial.

Contribution map

Human

  • Research question
  • Archival source material
  • Constraints and refusal
  • Selection and sequencing
  • Art direction
  • Ethical judgement
  • Written framing

AI

  • Generation
  • Variation
  • Speculative reconstruction

07

Output

An expanding body of images, texts and archival structures. Presentation format is in development.

08

AI disclosure

Why AI was used
The subject of the work is machine reconstruction itself; generative imagery is the only appropriate material.
What was verified
Historical and archival references underpinning the work are checked against sources; generated imagery is never presented as documentary.
What AI got wrong
Confident invention of period detail, and smoothing of ambiguity into resolved, legible scenes.
Rejected or changed
Output that read as convincing pseudo-documentary was removed or deliberately destabilised.
Human contribution
Concept, research, editorial selection, sequencing and all written material.
Ethics & authorship
Generated imagery is labelled as generated. The work does not present synthetic images as evidence of real events.

Skills demonstrated

  • Artistic research
  • Generative image practice
  • Archival thinking
  • Critical writing

Gallery

Speculative Archive / memory without lived experience — project plate.

Reflection

The work is most useful where the system fails: fluent, confident output that has no referent makes the gap between reconstruction and memory legible.

Related projects