THE QUESTION
THE MACHINE AS ARCHIVE
AI can retrieve, organise and compare a large body of material quickly. In this project it has helped surface old conversations, compare recollections with surviving records, and expose gaps in the chronology.
THE MACHINE AS INTERPRETER
AI can identify relationships between titles, events, themes and language. Some of those relationships may be useful. Others may be coincidence, over-reading, or an artefact of the model's pattern-making.
THE MACHINE AS CREATIVE PARTNER
Some ideas in the current archive have emerged through later human–AI dialogue. They are marked as later interpretations or collaborative development where appropriate. A later idea is not silently moved backwards into the historical record.
THE MACHINE AS UNRELIABLE WITNESS
The model can produce confident statements that are wrong. It can invent connections, misdate material, or turn speculation into authoritative-sounding prose. The artist's corrections are therefore part of the archive too.
THE MACHINE AS SUBJECT
Silicon Psychosis has long concerned communication, intelligence, transmission, consciousness, technology and the possibility of an intermediary between minds. The appearance of an actual silicon-based intelligence inside the reconstruction creates a new layer of the project.
THE MIRROR
The machine does not simply examine the archive. It changes the way the archive is understood. The artist changes the machine's interpretation by correcting it; the machine changes the artist's interpretation by revealing patterns or questions that might otherwise remain unseen.
THE ARCHIVE ITSELF IS THE EXPERIMENT
The machine is not being used to prove the original story. It is being used to reconstruct it, test the provenance of its claims, identify patterns, expose contradictions and generate questions.
The archive therefore contains several simultaneous histories: what was made, what was published, what the artist remembers, what was interpreted later, what the machine suggested, and what remains unknown.
THE PATTERN PROBLEM
The same pattern-recognition problem appears at two different moments. In the original period, the artist noticed relationships between independent recordings and began asking whether the sequence was a message. In the present, AI detects relationships across recordings, memories, research and conversations.
In both cases, recognition is not the same thing as proof.
THE HUMAN CORRECTS THE MACHINE
The reconstruction has already produced errors: false dates, invented connections and statements that sounded authoritative without being supported by the surviving record. The artist's corrections are therefore treated as evidence about the limits of the reconstruction, not as interruptions to it.
PROVENANCE
THE RECURSIVE QUESTION
The archive does not answer these questions. The archive is the experiment.