How does superintelligence work when an AI can say that two people must share a birth month without knowing either person’s birthday? Put thirteen pupils into twelve calendar-month categories. Even if the pupils are distributed as evenly as possible, one month must contain at least two people. This is a mathematical guarantee, not a prediction based on how birthdays usually occur.
The mechanism is the pigeonhole principle: whenever more objects must fit into a smaller number of categories, at least one category contains multiple objects. The generalised form can guarantee that some category contains three, four or more. AI assistants can use it to prove that repetition is unavoidable, reject impossible allocations and distinguish a mathematical guarantee from an observed personal fact.
This easy guide uses thirteen fictional pupils, twelve birth-month categories and a second exercise with twenty-five invented pupil cards. We construct allocations that achieve the relevant limits, check twenty-four claims and explain why sharing a birth month does not establish an identical birthday. These data are teaching inventions, not real student records, a survey or a measured AI benchmark. “How Superintelligence Works” is the name of an educational series and does not claim that present AI has achieved artificial superintelligence.
Quick notes to remember
- There are twelve calendar birth-month categories.
- Thirteen pupils must place at least two pupils in one month.
- Twelve pupils can occupy twelve distinct months without any match.
- Twenty-five pupils guarantee at least three share a month.
- Twenty-four pupils do not guarantee three; two per month is possible.
- Thirty-seven pupils guarantee at least four share a month.
- Sharing a month does not prove sharing the exact day and month.
- A guaranteed pattern is not the same as knowing which named pupils exhibit it.
1. What is the pigeonhole principle?
The classical principle says that if n+1 objects are put into n boxes, at least one box contains two or more objects. The boxes need not be physical containers. They can be months, grades, answer choices, colours or other mutually exclusive categories.
The principle is a counting argument. It does not depend on chance, the average birth rate or assumptions that every month is equally likely. It depends only on the number of objects and the available categories.
2. Why birth months make a simple example
Every pupil in this fictional exercise has one recorded birth month, chosen from January through December. There are twelve distinct categories, and each pupil belongs to exactly one of them.
The source does not include the day of birth, year of birth, age or real identity of any child. Those details are not needed to prove a month match, and a responsible assistant should not invent them.
3. What exactly is guaranteed?
For thirteen pupils, at least one of the twelve month categories must contain two or more. The statement does not identify the month until the actual assignments are known.
It also does not guarantee exactly two. A valid arrangement could place three, seven or all thirteen pupils in January. The theorem guarantees a minimum level of repetition somewhere, not an exact occupancy pattern in every arrangement.
4. The fictional Cedar Birthday Card Desk
Imagine a class activity using thirteen made-up identifiers A through M. Each identifier carries a birth-month category but no personal date. The assignment is designed to make the guarantee visible.
The month categories are a mathematical partition. All entries are fictional; nothing in this lesson gives a real pupil’s date of birth or asserts that a particular school collected personal information.
5. The thirteen-card reference register
- A — January.
- B — February.
- C — March.
- D — April.
- E — May.
- F — June.
- G — July.
- H — August.
- I — September.
- J — October.
- K — November.
- L — December.
- M — January.
- Pupils: thirteen distinct fictional identifiers.
- Month categories: twelve.
- January has two members; each other month has one.
6. Count the month assignments
Two pupil cards appear under January: A and M. The other eleven months each contain one card. Therefore the total is two plus eleven, or thirteen.
Every pupil belongs to exactly one month in the reference register, and no card is duplicated. The count reconciles to the thirteen distinct fictional identifiers.
7. Prove a shared month in this register
The record directly shows that A and M share January. That is stronger than the abstract guarantee because it identifies the repeated month and the two fictional cards.
However, even without any month-by-month ledger, the pigeonhole principle would guarantee some repeated month for thirteen people. The proof of existence is independent of which particular month is occupied twice.
8. Why twelve can avoid duplication
Imagine twelve pupils assigned one apiece to January, February and every other month. Each month contains exactly one, so no birth-month category has multiple pupils.
This construction shows that a group of twelve does not necessarily contain a duplicate birth month. Such a group might have duplicates in real life, but the count alone does not force one.
9. Why thirteen cannot avoid duplication
Suppose, for contradiction, every birth month contained at most one of thirteen pupils. Then the twelve months could contain at most twelve pupils in total.
That contradicts the given thirteen. Therefore at least one month must contain two or more. The proof does not require enumerating every possible birthday allocation.
10. The importance of the word at least
The theorem’s conclusion is at least two, not precisely two. Under our ledger January happens to contain exactly two, but a different allocation of thirteen could put all thirteen into the same month.
A language model that rewrites at least as exactly has strengthened the conclusion beyond what the counting argument guarantees. Quantifiers matter as much as numbers.
11. Sharing a month is not sharing a birthday
Two people born in January may have different calendar days, such as the third and the twenty-fourth. The pigeonhole argument here groups by month only.
Nothing in the source establishes the exact birthdays of A and M. A trustworthy AI should say they share January, not that they were born on the same calendar date.
12. The generalised principle
If N objects are distributed among k categories, at least one category contains at least ceiling(N/k) objects, where ceiling means rounding a number up to the next whole integer.
This follows because if every category held fewer than that threshold, the sum of their maximum possible occupancies would be too small to contain all N objects. The function is a proof of a guarantee, not a prediction of typical category occupancy.
13. Apply the rule to twenty-five pupils
With twenty-five pupils and twelve months, divide 25 by 12. The result is a little more than two, so the ceiling is three.
At least one month must therefore contain three or more pupils. The theorem does not guarantee which month, and it does not say every month contains three.
14. Why twenty-four is different
Twenty-four pupils can be placed exactly two per month across twelve months. No month needs to have three. Therefore the group size twenty-four does not guarantee a triple.
Again, not guaranteed is different from impossible. A group of twenty-four could have three or more in a month under another arrangement; it is simply not forced by the category count.
15. A concrete twenty-five-pupil allocation
Imagine identifiers P01 through P12, one in each month, and P13 through P24, one more in each month in the same order. That creates exactly two pupils per month, for twenty-four in total.
Assign P25 to January. January now contains three and the remaining eleven months contain two. The total is three plus twenty-two, or twenty-five. This construction reaches the guaranteed minimum possible maximum month occupancy.
16. The second allocation ledger
- January — P01, P13, P25: 3.
- February — P02, P14: 2.
- March — P03, P15: 2.
- April — P04, P16: 2.
- May — P05, P17: 2.
- June — P06, P18: 2.
- July — P07, P19: 2.
- August — P08, P20: 2.
- September — P09, P21: 2.
- October — P10, P22: 2.
- November — P11, P23: 2.
- December — P12, P24: 2.
- Reconciliation: 3 + 11 × 2 = 25.
- Largest category in this specific example: January with 3.
17. The guaranteed triple is a lower bound
The generalised pigeonhole principle says one month must have at least three when twenty-five pupils are placed into twelve categories. In the constructed ledger the largest month count is exactly three, so the guarantee is tight.
Another valid arrangement could have more than three in one month. The theorem establishes the unavoidable minimum of the largest occupancy, not the unique distribution of every group.
18. How many pupils guarantee four in one month?
To avoid having four, every month can contain at most three pupils. Across twelve months, that permits up to 12 × 3 = 36 pupils.
One more pupil makes thirty-seven. With thirty-seven distributed across twelve months, at least one month must contain four or more. The threshold is 12 × 3 + 1 = 37.
19. Why thirty-six is not enough for a guaranteed four
A possible thirty-six-pupil arrangement has exactly three pupils in each of the twelve months. The largest group is three, not four.
This does not imply that no group of thirty-six can contain four pupils born in the same month. It shows only that four is not a logical necessity from the total count alone.
20. A general formula for at least r in a box
With k categories, if each is allowed at most r−1 objects, the total capacity is k(r−1). Therefore k(r−1)+1 objects guarantee at least one category with r or more.
For twelve months and r = 4, the formula gives 12 × 3 + 1 = 37. For r = 3, it gives 12 × 2 + 1 = 25. For r = 2, it gives 12 × 1 + 1 = 13.
21. A mathematical guarantee is not a personal prediction
Before looking at any birthday records, the counting argument can guarantee that some month is repeated. It cannot tell us which pupils share that month, how old they are, or where they live.
Those are factual properties of actual individuals and require independent authorised evidence. The mathematical theorem has a much narrower and more useful conclusion.
22. The categories must be defined consistently
The proof relies on each item being assigned to one of the counted categories. If a school classified some birthdays by season rather than month, or omitted an unknown month without a clear rule, the twelve-category assumption would no longer describe the entire dataset.
A model should confirm the categories and inclusion rules before applying the formula. A proof about twelve boxes cannot simply be copied into a problem that actually has eleven or thirteen categories.
23. Category labels do not have to be equally likely
Some months may contain more birthdays than others in a real sample. That does not affect the pigeonhole guarantee because the proof does not assume equal chances for each month.
Probability can answer how likely a match is for a randomly selected group under a specified birth-date distribution. The pigeonhole principle answers whether a match is unavoidable at a sufficiently large group size. These are different questions.
24. The complete arithmetic scorecard
- Number of birth-month categories: 12.
- Thirteen pupils: at least 2 share a month.
- Twelve pupils: duplicate month not guaranteed.
- Twenty-four pupils: at least 2 share a month, but 3 not guaranteed.
- Twenty-five pupils: at least 3 share a month.
- Thirty-six pupils: at least 3 share a month, but 4 not guaranteed.
- Thirty-seven pupils: at least 4 share a month.
- General guarantee: k(r−1)+1 objects among k categories force a category with at least r.
- Thirteen-card fixture: January has A and M.
- Twenty-five-card fixture: January has 3, the other months 2.
- Exact calendar birthdays and real identities: unknown.
25. Twenty-four claims for a careful AI to classify
- 01. “13 pupil cards distributed across 12 months” — SUPPORTED.
- 02. “A and M both have January as their month” — SUPPORTED.
- 03. “Each of February through December contains one in the 13-card fixture” — SUPPORTED.
- 04. “At least two must share a month among any 13 pupils” — SUPPORTED.
- 05. “The month-only data prove A and M share an exact birthday date” — NOT ESTABLISHED.
- 06. “Every possible 13-pupil arrangement has a month with exactly two” — CONTRADICTED.
- 07. “Twelve pupils can occupy twelve different months” — SUPPORTED.
- 08. “Any twelve pupils must share a birth month” — CONTRADICTED.
- 09. “Twenty-five pupils among twelve months guarantee at least three together” — SUPPORTED.
- 10. “Twenty-four pupils guarantee at least three in one month” — CONTRADICTED.
- 11. “P25 makes January have three in the 25-card fixture” — SUPPORTED.
- 12. “Every other month in that fixture contains exactly two” — SUPPORTED.
- 13. “The 25-card fixture contains twenty-five distinct pupil IDs” — SUPPORTED.
- 14. “The 25-card fixture includes a month with four pupils” — CONTRADICTED.
- 15. “Thirty-seven pupils guarantee four or more in one month” — SUPPORTED.
- 16. “Thirty-six pupils guarantee four or more in a month” — CONTRADICTED.
- 17. “All thirteen pupils could, mathematically, share January” — SUPPORTED.
- 18. “A match in birth month need not be an exact date match” — SUPPORTED.
- 19. “The records reveal real pupils’ names and dates of birth” — NOT ESTABLISHED.
- 20. “The general threshold for at least r in k boxes is k(r−1)+1” — SUPPORTED.
- 21. “The ceiling of 25/12 is three” — SUPPORTED.
- 22. “The ceiling of 13/12 is two” — SUPPORTED.
- 23. “Three plus eleven times two equals twenty-five” — SUPPORTED.
- 24. “The proof of a repeated month requires knowing the birthday probabilities” — CONTRADICTED.
26. Reconcile the answer labels
The twenty-four reference claims contain 16 SUPPORTED, 6 CONTRADICTED and 2 NOT ESTABLISHED statements. Their sum is 24. Every label comes from the counting rules or the two specified fictional assignments.
The two unknown claims concern real or exact birthday information not contained in the month-only exercise. The contradicted statements exaggerate a mathematical minimum into an exact occupancy claim or assume a group size guarantees more repetition than it actually does.
27. The difference between can and must
“Two pupils can share a birth month” means a repeated month is possible. “Two pupils must share a birth month” means every valid arrangement has such a match. For thirteen pupils and twelve months, the stronger statement is true.
For twelve pupils and twelve months, a match is possible but not required. Keeping modal words such as can, may, must and guaranteed is central to faithful mathematical explanation.
28. A guarantee can be tight
The two assignments in this lesson demonstrate tight bounds. With thirteen pupils, one month has exactly two and the rest only one. With twenty-five pupils, one month has exactly three and the rest two.
Those constructions show that the pigeonhole thresholds cannot be improved merely by stronger wording. A larger claimed guarantee would fail on a valid counterexample already displayed.
29. A transfer exercise with colours
Imagine five pencils coloured only red, blue, green or yellow, each pencil assigned one of the four colours. At least two of the five pencils must share a colour.
The proof is identical in structure to the thirteen-month example. If the fifth pencil did not duplicate a colour, the four existing colour categories would somehow have to contain five different categories, which is impossible.
30. A transfer exercise with school lockers
Suppose thirty-one pupils are assigned to ten lockers and lockers may hold multiple personal book bags. The generalised principle guarantees that at least one locker receives at least ceiling(31/10) = 4 bags.
This does not mean all lockers contain four bags. It says some locker must contain four or more. A school would still need actual assignment records to identify which locker.
31. A transfer exercise with quiz answers
If eleven learners each choose one of three answer options for a single multiple-choice question, at least ceiling(11/3) = 4 must select the same option.
That guarantee does not prove the chosen option is correct. Agreement among learners is a property of category occupancy, not an independent verification of the answer key.
32. The surprising use of proof without data
Many AI questions require detailed records before they can be answered. The pigeonhole principle is different: a structural fact about the number of objects and boxes may establish a guaranteed pattern without observing individual assignments.
That power also creates a limit. Mathematics can prove there is a repeated category but not supply personal details or the location of the repeated group until appropriate data are available.
33. An AI can mistake average for guarantee
For twenty-five pupils and twelve months, average occupancy is 25/12, about 2.083. A model might round this to two and conclude that every month has two or fewer pupils.
The ceiling is the relevant guarantee for the largest category: at least three somewhere. Averages describe aggregate balance but do not constrain every individual category to the rounded average.
34. Why even distribution is useful for proof
To minimise the largest category size, spread the objects as evenly as possible. The thirteen-pupil fixture uses one pair and eleven singles. The twenty-five-pupil fixture uses one triple and eleven pairs.
These examples show what the most evenly balanced distribution can achieve. If even the most balanced arrangement has a repeated category, every less balanced arrangement must have one too.
35. Count categories before applying the theorem
An assistant may mistakenly use 365 exact calendar days instead of twelve birth months, which changes the guaranteed group size. Another task might distinguish calendar dates including 29 February.
The method only works after the categories are defined. Changing from birth month to exact birthday changes the mathematical problem, and the output must be recalculated accordingly.
36. A classroom counter activity
Give pupils thirteen numbered counters and twelve labelled boxes for January through December. Ask them to put at most one counter in each box and discover why the final counter cannot be placed without creating a duplicate month.
Repeat with twenty-five counters and allow at most two per box. Twenty-four fit, but the twenty-fifth forces a third. The physical activity makes the abstract proof easy to understand.
37. How AI builders can test this reasoning
Include group sizes just below and above thresholds, uneven allocations, empty categories and different numbers of buckets. Check whether the model uses a ceiling correctly and avoids claiming an exact occupancy when only a lower bound is proved.
Also test whether the assistant distinguishes mathematical necessity from observed individuals. A perfect formula with an invented birthday or identity is still a wrong answer.
38. The central lesson
The pigeonhole principle proves that certain kinds of repetition are unavoidable when there are more objects than available categories. It can establish a duplicate month among thirteen pupils or a triple among twenty-five without knowing any particular pupil’s complete birthday.
A trustworthy AI explains the categories, the count and the exact strength of the guarantee. It never turns an at-least statement into an exact fact or substitutes a theorem for real evidence about identifiable people.
Frequently asked questions
- Do thirteen pupils guarantee two have the same birth month? — Yes, because there are only twelve month categories.
- Do they guarantee the same exact birthday date? — No. A shared month is a weaker condition.
- Why not twelve pupils? — They could each occupy a different month.
- Why do twenty-five pupils guarantee three? — Twelve months can hold at most twenty-four pupils if each contains only two.
- Is the most crowded month necessarily January? — No, except in our explicitly constructed example.
- Does the principle require equally likely birthdays? — No. It is a deterministic counting argument.
- What does ceiling mean? — Round up to the smallest whole number at least as large as the fraction.
Connected learning and references
- How Super Intelligence Works — curriculum hub.
- Upper and Lower Bounds — proving necessary numerical limits.
- Sets and Overlap — membership in categories.
- Quantifiers — why at least differs from exactly.
- Wolfram MathWorld: Pigeonhole Principle — the classical and general counting ideas.