☯ Ming Luo Pan

Essays › casting method · 2026-10-20

Is AI Fortune-Telling Accurate? — First Separate "Casting" from "Reading"

"Is ChatGPT fortune-telling accurate?" "Does AI Zi Wei Dou Shu work?" — these searches keep climbing, but most answers blur two different jobs into one. Accuracy depends entirely on which layer the AI is doing.

Short version: casting a chart is math; interpreting it is language. AI can take over the second safely. Ask it to do the first, and you get the problem below.

Two layers, two different questions

Every metaphysics reading splits into two steps:

  1. Chart casting (deterministic) — birth date/time in, ganzhi, stars and palaces out, by fixed calendrical rules. This step has one right answer: a given timestamp produces a given chart, computed not guessed.
  2. Interpretation (probabilistic) — translating chart symbols into plain language. What does "Tian Ji transforming to Ji in the Parents palace" mean for this person? No single answer; quality depends on interpretive depth.

So the right question isn't "is AI fortune-telling accurate" but "which layer is this site's AI responsible for?"

The trap of asking ChatGPT to cast a chart: it invents ganzhi

Ask a general LLM to "cast my Zi Wei chart" and it produces a beautifully formatted, confident-looking chart — with palace placements that are frequently wrong.

The reason is simple: LLMs are language models. They "remember" roughly what a Zi Wei chart looks like, but star placement has dozens of rules (Five-Element Bureau locates Zi Wei, Zi Wei anchors the northern and southern star groups, four transformations fly across palaces...). Probabilistic completion slips on one rule and the whole chart goes wrong — and worse, it fails confidently. It won't say "actually, I guessed."

The same applies to any service where the AI generates the chart directly: pretty layout, wrong math.

The right division of labor: engine casts, AI reads

Serious implementations split the work:

The first layer guarantees no errors; the second plays to AI's real strength — organizing scattered knowledge into language that's useful to you.

Three questions to test an AI fortune-telling site

Next time you see an AI-branded metaphysics site, ask:

  1. Can the chart be verified? — trustworthy sites show the full star/stem/palace table so you can cross-check against other software or books. If it jumps straight to conclusions without showing data, be careful.
  2. Did the AI receive actual chart data before interpreting? — if it "imagines" a chart from your birthdate, the error rate is high.
  3. Do predictions leave a verification trail? — serious sites timestamp their public calls and mark hits and misses when they come due — not cherry-pick winners afterward.

A site that passes all three is worth your time.

Common misconceptions

How this site does it

Ming Luo Pan is built on exactly this "engine casts, AI reads" architecture:

Verify it yourself: cast a free Zi Wei chart and check the star table against any reference book, or cast your BaZi and inspect the ten-gods layout — every cell is traceable.

Traditional divination is a cultural reference and self-exploration tool — not medical, legal, or investment advice.