Source Attribution: Why Transparency Matters More Than Being Right

The Trust Problem with AI Readings

When an AI hands you a reading about yourself, how do you know it is not simply “making things up”? Large language models are very good at producing text that sounds plausible but has no basis in fact—which is especially dangerous in the realms of fortune-telling and psychology, where users often lack the ability to judge whether professional-sounding content is accurate.

Traditional fortune-telling apps typically deliver conclusions without their reasoning: “Your finances will be poor this year,” “Your marriage will hit rough patches”—there is no way to verify how these conclusions were derived; you can only choose to believe them or not.

Astralis’s Source Attribution System

Astralis attributes every AI reading along three dimensions: the source system (BaZi / astrology / psychology / integration), the corpus identifier (the specific data field or rule), and the confidence level (high / medium / low / derived).

For example, a reading such as “Your Day Master is Ding Fire, born in the Wu month when Fire is at its height” is attributed to “BaZi · Day Master” with the corpus identifier “day_master_strength.” You can trace it back to the Day Stem and Month Branch in the BaZi chart data and verify the basis of the reading yourself.

What the Confidence Levels Mean

• High confidence: a direct reading from the chart data, or a signal on which all three systems agree
• Medium confidence: an indirect inference from the chart’s configuration, or a signal on which two systems agree
• Low confidence (derived): psychological traits AI inferred from the natal chart; we recommend verifying with an actual assessment
• To verify: areas where the three systems differ; we recommend you observe them in your daily life

Transparency Matters More Than Being Right

We do not claim Astralis’s readings are necessarily “correct”—neither fortune-telling nor psychology is an exact science, and any reading can be inaccurate. What we do guarantee is that every reading is “traceable”—you know what it is based on and can judge and verify it yourself.

This transparency turns you from a “passive receiver of conclusions” into an “active participant in the interpretation.” You can see which data the AI used, which rules it applied, and which conclusions it drew—and if you disagree, you can examine the raw data and form your own judgment. That is what a genuine tool for self-knowledge should provide.