Asking a General AI to Cast a Qi Men Chart vs Using a Charting Program: Three Examples You Can Check

Published

Many people open a general chat AI such as ChatGPT, Gemini or DeepSeek and type “cast a Qi Men chart for right now”. The answer usually looks tidy and full of the right terms. But casting a Qi Men chart is not writing; it is calendar math with one right answer. This article explains where the two approaches differ and shows the difference with three examples computed by our program, which you can check yourself.

1. The basic difference

Asking a general AI to cast the chartA program casts it, the AI only explains
Where the chart comes fromText written by probability, as if “writing” a chartFixed rules, computed step by step
Solar-term boundariesOften only “around February 4”, not the minuteComputed to the minute from the sun’s ecliptic longitude
True Solar Time, daylight savingNot always applied; easy to miss for places outside ChinaConverted from the place’s time-zone rules and longitude
23:00 day changeNot always appliedThe day changes at 23:00 True Solar Time
Asking twiceThe chart may differAlways the same
Can you check it?Hard to see how it got thereEvery palace can be opened and checked; the rules are public
What the reading rests onIf the chart is wrong, a fluent reading explains a different chartThe AI gets the computed chart and cannot change it

2. Three examples you can check

Every chart below was computed by Qimena’s charting engine (hour-based Qi Men, Chai Bu to set the ju, the rotating plate to lay it out); the rules are on How Qimena casts a chart. Each example compares the correct conversion with one step missed.

Example 1: at a solar-term boundary, half an hour means another year

Lichun (Start of Spring) 2027 falls at 09:46 on February 4, Beijing time. Charts cast in Shanghai:

Time (Shanghai)Year pillarMonth pillarJuZhi FuZhi Shi
2027-02-04 09:30丙午辛丑Yang Dun 9Tian YingScenery door
2027-02-04 10:00丁未壬寅Yang Dun 5Tian QinDeath door

Thirty minutes apart, and the year pillar, month pillar, ju, Zhi Fu and Zhi Shi all differ; the Open door moves from Kan (north) to Zhen (east). An answer that only knows “Lichun is on February 4” is wrong for half of that morning.

Example 2: noon in New York under daylight saving, and the best direction moves

July 15, 2026, 12:00 on a New York clock (daylight saving time). Remove daylight saving, then correct for longitude (about 74° W) and the equation of time, and True Solar Time is about 10:58.

MethodHour pillarOpen doorLife door
Converted to True Solar Time (correct)辛巳Xun 4 (southeast)Kun 2 (southwest)
Cast straight from 12:00壬午Gen 8 (northeast)Xun 4 (southeast)

The ju (Yin Dun 5) happens to match, but the hour changes from Si to Wu and the eight doors rotate: for a career question you look at the Open door, and the right answer is southeast; miss the conversion and it becomes northeast.

Example 3: 11:30 at night in New York, and even the day changes

Same day, 23:30 on a New York clock (daylight saving), True Solar Time about 22:28, still the Hai hour.

MethodDay pillarHour pillarOpen door
Converted to True Solar Time (correct)庚寅丁亥Gen 8 (northeast)
Cast straight from 23:30辛卯戊子Dui 7 (west)

Going by the clock, 23:30 is past 23:00, so the Zi-start rule makes it the Zi hour of the next day: the day and hour pillars are both wrong, and the Open door jumps from northeast to west. Readers outside China often check charts in the evening, so this is the most common slip.

3. Where a general AI tends to go wrong

Unless it calls a reliable charting program, a general chat AI can slip on any of these:

  • Solar terms by date, not by minute: on a boundary day the chart is wrong for part of the day (Example 1).
  • No daylight-saving removal, no longitude or equation-of-time correction: most common for places outside China (Examples 2 and 3).
  • Inconsistent day change: some answers change the day at midnight, some at 23:00.
  • Mixed tables and schools: different methods of setting the ju, rotating with flying plate; mixed together, the chart belongs to no school.
  • Different answers each time: each answer is generated afresh, so the ju or the Zhi Fu can change.
  • “Classical quotes” that do not exist: language models complete sentences that sound right.

None of this means a general AI is useless. For terms, concepts and sorting out your thoughts it is handy. But the casting step belongs to a program that follows the rules.

4. Test it yourself (checklist)

Send these three times to any general AI exactly as written. Ask for the year, month, day and hour pillars, Yin or Yang Dun and the ju number, the Zhi Fu, the Zhi Shi, and the palace of the Open door. Then compare:

  1. Shanghai, February 4, 2027, 09:30 (correct: 丙午 year, 辛丑 month, Yang Dun 9, Zhi Fu Tian Ying, Zhi Shi Scenery door, Open door in Kan 1).
  2. New York, July 15, 2026, 12:00 daylight saving time (correct: True Solar Time about 10:58, 辛巳 hour, Yin Dun 5, Zhi Fu Tian Fu, Zhi Shi Block door, Open door in Xun 4).
  3. New York, July 15, 2026, 23:30 daylight saving time (correct: 庚寅 day, 丁亥 hour, Yin Dun 5, Zhi Fu Tian Chong, Zhi Shi Harm door, Open door in Gen 8).

You can also enter the same time and place in the Qi Men calculator and open each palace to check.

5. How “the program casts, the AI explains” works

Qimena works in two layers:

  1. The program casts the chart: clock time is converted to True Solar Time from the place’s time-zone rules (daylight saving included), longitude and the equation of time; solar terms are computed to the minute from the sun’s ecliptic longitude; Chai Bu sets the ju and the rotating plate lays it out, and every palace can be opened and checked.
  2. The AI explains it: the AI gets the computed chart, each palace’s score, the patterns found, and time candidates the program picked by casting every shichen (two-hour period) of the next 7 days. It must tie each suggestion to a specific palace and door, cannot change the chart, and cannot name dates outside the candidates.

The full rules, what the AI can and cannot do, and why another app may differ are on How we calculate. For more on how AI readings work, see Are AI readings reliable?

Every chart value in this article was computed by the charting program and can be checked in the calculator; the article contains no test results for any specific AI product. Readings are for cultural study and entertainment only.

FAQ

Is a chart cast by a general AI always wrong?

Not always. If it calls a reliable charting program, or your time is far from any solar-term or hour boundary, it may be right. The trouble is you can rarely tell whether it calculated or just wrote, so check each item: the year, month and hour pillars, Yin or Yang Dun and the ju number, the Zhi Fu, the Zhi Shi, and which palaces hold the Open and Life doors.

Why does a general AI give a different chart when I ask twice?

A general chat AI writes its answer afresh each time, with some randomness; without fixed calculation steps, the ju or the Zhi Fu can change between answers. A charting program does not: the same time and place always give the same chart.

How is the AI in Qimena different from asking a general AI?

Qimena first casts the chart with a program using fixed rules (Chai Bu to set the ju, the rotating plate to lay it out, with True Solar Time and daylight saving handled), then gives the computed chart to the AI to explain. The AI cannot change the chart or name dates outside the candidates the program found. See our How we calculate page.