AI and Computer-Generated Astrology Disclosure: A Source-Reviewed Guide
Disclosure identifies which layer produced each visible claim. Scope and limits are explicit.
Overview
The interface distinguishes deterministic chart calculation, authored lookup content, and AI-generated narrative, while exposing model limits, uncertainty, review status, and correction paths. The page shows exactly what to verify, how to repeat the method, where a plausible near miss fails, which variants change the result, and what the evidence cannot establish.
At a glance
- Direct scope
- Reviewed finding — A chart pipeline contains distinct layers: deterministic calculation, curated lookup material, retrieval, model-generated narrative, human review, and user edits. Disclosure identifies which layer produced each visible claim.
- Evidence to verify
- Method checkpoint — Attach provenance to every report: calculator and ephemeris version, source record IDs, model and prompt version, generation time, safety interventions, reviewer status, confidence note, and a correction or deletion control.
- Repeatable method
- Worked-case result — A planet longitude is labelled “calculated,” its traditional meaning “editor-reviewed lookup,” and a personalized paragraph “AI-generated, unreviewed.” Opening the receipt reveals inputs and sources for each layer.
- Worked example
- Failure condition — A small “powered by AI” footer is insufficient when fluent prose appears authoritative, invents chart placements, or merges source text with model inference. Disclosure must sit beside the affected output.
- Near miss
- Documented variant — Some systems use templates without a generative model; others retrieve sources before generation or require human approval. The label should describe the actual pipeline rather than using “AI” as one undifferentiated category.
- Limits and safety
- Use boundary — Disclosure does not cure fabricated evidence, privacy violations, discrimination, or unsafe advice. Generated content remains subordinate to verified chart facts, consent, professional boundaries, and an accessible human correction route.
Definition and Scope
The interface distinguishes deterministic chart calculation, authored lookup content, and AI-generated narrative, while exposing model limits, uncertainty, review status, and correction paths.
A chart pipeline contains distinct layers: deterministic calculation, curated lookup material, retrieval, model-generated narrative, human review, and user edits. Disclosure identifies which layer produced each visible claim.
Evidence and Repeatable Method
Attach provenance to every report: calculator and ephemeris version, source record IDs, model and prompt version, generation time, safety interventions, reviewer status, confidence note, and a correction or deletion control.
Worked Example and Near Miss
A planet longitude is labelled “calculated,” its traditional meaning “editor-reviewed lookup,” and a personalized paragraph “AI-generated, unreviewed.” Opening the receipt reveals inputs and sources for each layer.
A small “powered by AI” footer is insufficient when fluent prose appears authoritative, invents chart placements, or merges source text with model inference. Disclosure must sit beside the affected output.
Variants and Disagreement
Some systems use templates without a generative model; others retrieve sources before generation or require human approval. The label should describe the actual pipeline rather than using “AI” as one undifferentiated category.
Limits and Safe Use
Disclosure does not cure fabricated evidence, privacy violations, discrimination, or unsafe advice. Generated content remains subordinate to verified chart facts, consent, professional boundaries, and an accessible human correction route.
Sources and editorial basis
- NIST AI Risk Management Framework 1.0NIST AI 100-1, section 2 “Audience”, section 3 “AI Risks and Trustworthiness”, and Part 2 “Core and Profiles”, pages 4–32A public framework for valid, reliable, transparent, explainable, privacy-enhanced, fair, and accountable AI risk management.
- NCGR Code of EthicsNCGR Code of Ethics, sections A.2–A.4, B.1–B.3, C, D, E, and F on harm, competence, confidentiality, public claims, and complaintsA public professional code addressing harm, competence, qualified interpretations, confidentiality, privacy, advertising, research, and complaint handling.