Reading generated images as symbolic material surfaced by a system under pressure. Tech divination treats the image as an encounter rather than decoration: ask clearly, generate honestly, then notice both what obeyed the prompt and what did not.
Prompt Engineering Application
Shape What it is
- Live question — begin with something genuinely open.
- Literal reading first — figures, omissions, and distortions before meaning.
- Split reading — what obeyed the prompt is separated from what did not.
Test: If the image was only decoration for a conclusion already held, it was not divination.
Motion How it moves
Ask → Generate → Read Literally → Interpret Together
- Ask — pose a real question without prescribing its answer.
- Generate — make the image through an honest pipeline.
- Read — take the literal content first, including absences.
- Interpret — follow charged details together over time.
Recursions The same pattern at two scales
Close
- Single reading — one image answering one question.
- Prompted versus unbidden — separating what was asked from what arrived.
- Charged detail — a small element that keeps drawing attention.
Vast
- Tarot and lots — older traditions of reading generated symbols.
- Scrying — seeking sight in a medium rather than a message.
- Oracle traditions — questions put to a system that answers obliquely.
Ethics What it refuses
- Prophecy — presenting a reading as a fixed outcome.
- Outsourcing choice — letting the image decide what a person should do.
- Laundering — turning a refusal into a flattering result and calling it the answer.
Read for possibility, not verdict. The image shows; it does not decide.
Practices
- Set the question — decide what is being asked before generating.
- Read the literal — describe the image plainly before interpreting.
- Follow the charge — return to recurrent details across a sequence.
- Let time answer — let later events test the reading.