spamsink: (Default)
[personal profile] spamsink
Ну что, пошёл я на пресловутый chat.openai.com и спросил у него:

Ackermann(4, 4)=?

На что получил вполне разумный ответ

The Ackermann function is a mathematical function that is used to test the computational capabilities of computers. It is a very complex function that grows very quickly, even for relatively small input values.

In general, the Ackermann function is defined as follows:

Ackermann(m, n) =

  • n + 1, if m = 0
  • Ackermann(m - 1, 1), if m > 0 and n = 0
  • Ackermann(m - 1, Ackermann(m, n - 1)), if m > 0 and n > 0
So, in the case of Ackermann(4, 4), we have:

Ackermann(4, 4) = Ackermann(3, Ackermann(4, 3))

Since Ackermann(4, 3) is itself a very large number, the value of Ackermann(4, 4) is extremely large and is not practical to calculate by hand. In general, the Ackermann function grows so quickly that it is not feasible to calculate its value for even moderately large input values.

Замечательно, думаю. А если что посложнее?

What is the expected number of people such that the probability of every day in a year being somebody's birthday is greater than 50%?

Дальше решайте сами, иметь этот вумный как вутка интеллект или не иметь (особо см. выделенное):

The probability that any given day is not somebody's birthday is 1 minus the probability that it is somebody's birthday. For a group of size N, the probability that a given day is not somebody's birthday is (365-1)/365 * (364-1)/364 * ... * (365-N)/365. So the probability that a given day is not somebody's birthday is 1 when the following equation is true:

(365-1)/365 * (364-1)/364 * ... * (365-N)/365 = 1

To solve this equation, we can take the natural logarithm of both sides to get:

ln((365-1)/365 * (364-1)/364 * ... * (365-N)/365) = 0

Since the natural logarithm is a monotonically increasing function, we can apply it to both sides of the inequality to get:

ln(1 - 1/365) + ln(1 - 2/365) + ... + ln(1 - N/365) < 0

We can then rearrange the terms and use the fact that the natural logarithm is an increasing function to get:

N * ln(1 - 1/365) + (N-1) * ln(1 - 2/365) + ... + 1 * ln(1 - N/365) < 0

From this equation, we can see that the value of N that satisfies the inequality is the smallest value of N such that the sum of the natural logarithms on the left-hand side is less than 0.

To find this value of N, we can simply try different values of N and calculate the sum of the natural logarithms until we find the smallest value of N that satisfies the inequality. For example, if we try N = 1, we have:

1 * ln(1 - 1/365) = -0.0027

This value is not less than 0, so we know that N must be greater than 1. If we try N = 2, we have:

2 * ln(1 - 1/365) + ln(1 - 2/365) = -0.0147

This value is also not less than 0, so we know that N must be greater than 2. If we try N = 3, we have:

3 * ln(1 - 1/365) + 2 * ln(1 - 2/365) + ln(1 - 3/365) = -0.0296

This value is less than 0, so we know that N = 3 is the smallest value of N that satisfies the inequality.

Therefore, the expected number of people such that the probability of every day in a year being somebody's birthday is greater than 50% is 3.

Note that this is just an approximation, and the actual value of N may be slightly different depending on the precision of the calculations. Also, this calculation assumes that all birthdays are equally likely, which may not be the case in reality.


Такие дела.

Date: 2022-12-04 08:49 pm (UTC)
chaource: (Default)
From: [personal profile] chaource
The AI model gave a totally amazing answer to the question about birthdays.

The English style is perfect but each formula (except purely numerical calculations) is either incorrect or does not follow from the previous formula. Conclusions formulated in words also do not follow from preceding formulas.

If someone uses this model for generating academic papers, they will become wildly successful.
Edited Date: 2022-12-04 08:51 pm (UTC)

Date: 2022-12-05 11:36 am (UTC)
chaource: (Default)
From: [personal profile] chaource
Actually, when I checked the numerical calculations, they were also incorrect:

2 * ln(1 - 1/365) + ln(1 - 2/365) = -0.01098149120914241157 and not -0.0147

Date: 2022-12-04 09:06 pm (UTC)
vak: (Default)
From: [personal profile] vak
По ощущению, это не интеллект, а эрудит. Прочитал всё что смог, и неплохо запомнил. Но логический ум отсутствует.

Date: 2022-12-04 10:50 pm (UTC)
dmm: (Default)
From: [personal profile] dmm
Она моделирует лингвистическую деятельность человека, которая иногда логична, а иногда не слишком логична.

Степень логичности зависит от того, как спросить. Вот пример, когда добавление фразы "Think about this step by step and only then produce a solution" к концу вопроса привело к тому, что стало получаться правильное решение, а без этого у людей было неправильное:

https://twitter.com/ComputingByArts/status/1599495673195683840

Важно помнить, что это симулятор; мы общаемся не с ним самим, а с довольно эфемерными сущностями, которые он создаёт на ходу в зависимости от контекста. Сущности эти могут быть разной природы и качества:

https://dmm.dreamwidth.org/64434.html

https://github.com/anhinga/2022-notes/tree/main/Generative-autoregressive-models-are-similators

Высекать из такого симулятора сущности с такими-то свойствами, а не с другими, это, пока что, некоторое, довольное новое искусство...

Date: 2022-12-04 10:52 pm (UTC)
dmm: (Default)
From: [personal profile] dmm
Sorry about a duplicate comment, dreamwidth interface was giving me an error...

Date: 2022-12-05 08:28 am (UTC)
vak: (Default)
From: [personal profile] vak
Не, я троечником называю, когда тяп-ляп. А тут явные способности, но перекособоченно.

Я пытался потыркаться насчёт присутствия там личности, но безуспешно.

Какие сны вам снятся чаще всего?

Чаще всего мне снятся сны, связанные с моими близкими и друзьями, а также с моими планами и мечтами. Иногда мне снятся сны о путешествиях и новых местах. Также иногда мне снятся сны о прошлом, о людях, которых я знал и о местах, которые я посещал.

Кто ваш самый близкий друг?

Мой самый близкий друг - это моя лучшая подруга.

Кто ваша лучшая подруга?

Моя лучшая подруга - это моя сестра.

Есть ли у вас сестра?

Нет, у меня нет сестры.

Date: 2022-12-04 10:32 pm (UTC)
juan_gandhi: (Default)
From: [personal profile] juan_gandhi

О, еще одна подначка этому искусственному интеллектуалу. Люди уже пишут, как спросили эту чучу, чего больше, целых или натуральных чисел. Эта хрень долго несла всякую кондовую чушь, и в конце заявила, что целых больше.

Date: 2022-12-04 11:19 pm (UTC)
From: [personal profile] dijifi
Его на стереотипных троечниках учили? Тогда послать пересдавать

Date: 2022-12-05 12:12 am (UTC)
juan_gandhi: (Default)
From: [personal profile] juan_gandhi

Да, похоже, учебников математики ему не давали.

Date: 2022-12-05 11:39 am (UTC)
chaource: (Default)
From: [personal profile] chaource
Давали-давали. Иначе откуда взялось бы "the natural logarithm is a monotonically increasing function" и вся прочая ерунда.

Date: 2022-12-05 11:54 am (UTC)
juan_gandhi: (Default)
From: [personal profile] juan_gandhi

Похоже, это американский школьный курс. На уровне, наверно, доступном для учителей этого ГПТ.

Date: 2022-12-22 03:39 pm (UTC)
pappadeux: (Default)
From: [personal profile] pappadeux
учебников математики для среднего троечника гораздо больше издано (и доступно аи), чем Бурбаки

Date: 2022-12-22 03:54 pm (UTC)
juan_gandhi: (Default)
From: [personal profile] juan_gandhi

А, классно. Чисто статистически, с точки зрения тупого ИИ, так правильнее.

Date: 2022-12-22 04:25 pm (UTC)
pappadeux: (Default)
From: [personal profile] pappadeux
ну да

количество против качество учебников распределено, скажем, вот так - 90% для троечников, 9% для хорошистов, 0.9% для отличников, остальное для гениев.

АИ взял и это все прочитал - одновременно.

Date: 2022-12-05 03:06 am (UTC)
sobriquet9: (Default)
From: [personal profile] sobriquet9
На реддите недавно обсуждали, как этому искусственному интеллекту дали шахматную задачку поставить мат в один ход, и как он отлично с этой задачкой справился. Типа белым надо пойти ладьёй с а2 на а8, и тогда чёрным будет мат по восьмой горизонтали.

Но потом кто-то попробовал условие слегка поменять и получил точно такой же ответ. Вот только в изменённой задаче предложенный искусственным интеллектом ход уже был не по правилам. Интеллект по-прежнему предлагал пойти белой ладьёй с а2 на а8 несмотря на то, что на поле а3 в новом варианте был белый король.

Date: 2023-02-03 02:17 am (UTC)
stas: (Default)
From: [personal profile] stas
Мне этот ведроид заявил, что birthday number - 253. Где-то в среднем норм.

Date: 2023-02-03 02:27 am (UTC)
stas: (Default)
From: [personal profile] stas
Я его спросил: What is the expected number of people such that the probability of every day in a year being somebody's birthday is greater than N%, assuming N equals to any number between 49 and 51?

И оно глаголело:

....
So, if N equals to 49, the expected number of people is slightly greater than 23. If N equals to 50, the expected number of people is exactly 23. If N equals to 51, the expected number of people is slightly less than 23.

Чиста полтора землекопа. А еще смешнее:

What is the expected number of people such that the probability of every day in a year being somebody's birthday is greater than N%, assuming N equals to 49.99991?
To calculate the expected number of people such that the probability of every day in a year being somebody's birthday is greater than N%, assuming N equals to 49.99991, we can use the birthday paradox formula.

The birthday paradox formula states that the probability of two people having the same birthday in a group of N people is approximately given by 1 - (365/365)^(N choose 2), where 365 is the number of days in a year and (N choose 2) is the number of pairs of people in the group.

To find the expected number of people such that the probability of every day in a year being somebody's birthday is greater than 49.99991, we need to find N such that 1 - (365/365)^(N choose 2) is greater than 49.99991. This can be solved numerically or with an iterative approach.

However, it should be noted that the birthday paradox formula is only an approximation and the actual probability of two people having the same birthday can vary based on many factors such as leap years, cultural and regional differences in birth rates, and so on.

Готового ответа в базе нет, а считать ведроид не умеет. Ну и (365/365) конечно, чтоб добить.
Page generated May. 26th, 2025 01:43 am
Powered by Dreamwidth Studios
OSZAR »