
Kep, Cambodia · November 2026
THREE CHANNELS
The Unasked Cross
One sequence of found objects, chance, and Tarot was submitted to different artificial intelligences. Each AI selected its own material evidence, constructed a symbolic argument and produced or proposed a visual artwork.
01 / The Encounter
Three inputs. No question.
Within approximately ten minutes, unrelated objects were arranged on white paper, three dice were rolled, and a Tarot Celtic Cross was laid without a question. The resulting photographs became a shared point of departure.
02 / The Procedure
Identical inputs. Independent decisions.
The same photographs and instructions were given to several AI models. Each selected three objects, interpreted the Tarot spread through its traditional symbolism, extracted four visual operations and imposed two irreversible eliminations.
3
Selected objects
4
Symbolic operations
2
Irreversible eliminations
The Tarot reading served as a symbolic input. The experiment makes no claim about prediction and examines the forms of meaning artificial intelligences construct from the same ambiguous encounter.
03 / The Results
Independent interpretations.
The selected objects and visual procedures differed, but openings, fractures and directional movement recurred across several results. The models used different image-making workflows.
Grok · Score only
An image not made
Grok selected the egg, tape roll and banana and developed a four-operation score. Its initial run produced no image.
After the first interpretations
ChatGPT later executed Grok’s unrealized score, producing a separate interpretation of its prescribed operations.
04 / The Observation
Different objects. A recurring opening.
Despite independent choices, several artworks return to a common geometry. An enclosure breaks, and something moves beyond it. The repetition may reflect the Tarot reading, the visual properties of the selected objects or conventions learned by AI systems. The experiment does not establish which influence was decisive.
05 / The Question
When artificial intelligences give form to the same symbolic encounter, are they discovering something about the encounter or revealing how they have learned to construct meaning?
Xgeronimo + AI
Original encounter and source photographs by Xgeronimo. AI interpretations and renderings are credited individually. This artistic comparison used different workflows and did not control for differences between models.
← Return to Experiments