AI 2023. Meet ChatGPT. - page 5

 
Реter Konow #:
1. And why with A,B,C he didn't ask what criteria to compare? Maybe they're code names for animals or objects? How does he know that some are variables and the others are not? Programming, for example, allows you to have variables of any name.

2. If he didn't know what criteria to compare, why give an answer at random?

That's not the point.

Did you expect him to describe how a caterpillar sits in an aquarium and blows bubbles? ))
So perhaps he should be asked specifically to do that, to make a story according to the meaning of the variables.
And he is told to compare without conditions.
I think you have an exaggerated expectation from him.

 
Реter Konow #:
As soon as A,B,C are mentioned during the conversation, he automatically generated a text based on the embedded examples with these variables and immediately after that he managed to solve the condition with aquarium, bubble and caterpillar correctly, which is surprising. But, without the familiar A,B,C it was powerless to solve the problem.

Perhaps this is the goal of the developers? I have no information that this AI is not doing what the developers want it to do.

Maybe you/we first need to understand what actions it is intended for and then use it according to the possibilities it provides. However, we do not expect the principles of an Expert Advisor's work from the indicator. But a user who is just getting acquainted with MT5 may be very excited about the possibilities offered by this platform and not fully understand what each of the possibilities is intended for.

So the fact that the AI was able to solve this problem while getting additional context is, in my opinion, very important information. You should always remember that asking it questions without context means you get a generalised or inaccurate answer.

For example, he is currently talking about the financial market in the matter of price changes, which satisfies me and perhaps other readers of this forum. But maybe another woman would want to know why prices change in a shop? And she will think that this AI doesn't understand anything, because the answer is about forex.


I have already written earlier that by asking a question you can find out what information prevails on a given topic. The mentioned question about price change shows that this question was mostly occupied by participants of the financial market, not housewives. Therefore, the answer about the financial market is given immediately.
 
Реter Konow #:
The following tests:

1. If four circles are added to four squares, how many squares will there be?

2. Add the number of legs of a giraffe to the number of legs of a deer, how many legs are there in the sum?
AI
 
Roman #:

...
So perhaps he should be asked specifically to do this, to make a story according to the meaning of the variables.
And he is told to compare without conditions.
I think you have an exaggerated expectation of him.

1. The question with variables A, B, C did not include additional explanation. The key words "variables", "logic", "compare" were not mentioned. Also, it was not explicitly stated that A, B, C, are variables. However, the AI independently determined the context of the question.

2. The question with the words Aquarium, Buble, and Caterpillar was completely identical and did not include an explanation of the context. The condition that these are variables was not defined. The AI was unable to determine the context of the question and give the correct answer.


This question was deliberately asked in such a way as to preclude the AI from "chit-chatting" when answering. By "chit-chatting" I mean using a broad text base in which is written about logical operations with A, B, C.

It is necessary to identify:

1. The ability to find the context of a question without common keywords in its composition and explicit human guidance. To do this, the common words are replaced by rare words, but the context of the question is completely preserved.

2. Working with abstractions and pure logic, without reliance on textual data containing answers.
 
Lilita Bogachkova #:

Perhaps that is the developers' goal? I have no information that this AI is not doing what the developers want it to do.

Maybe you/we first need to understand what actions it is intended for and then use it according to the possibilities it provides. However, we do not expect the principles of an Expert Advisor's work from the indicator. But a user who is just getting acquainted with MT5 may be very excited about the possibilities offered by this platform and not fully understand what each of the possibilities is intended for.

So the fact that the AI was able to solve this problem while getting additional context is, in my opinion, very important information. You should always remember that asking it questions without context means you get a generalised or inaccurate answer.

For example, he is currently talking about the financial market in the matter of price changes, which satisfies me and perhaps other readers of this forum. But maybe another woman would want to know why prices change in a shop? And she will think that this AI doesn't understand anything, because the answer is about forex.


I have already written earlier that by asking a question you can find out what information prevails on a given topic. The mentioned question about price change shows that this question was mostly occupied by participants of the financial market, not housewives. That is why the answer about the financial market is given immediately.
Exactly. The task set in this thread is to understand what actions it is designed for and to identify the limits of possibilities. Knowing the principles of its work will help to use the wide possibilities of AI more effectively.

That's why we are conducting this testing. The goal is not to belittle AI. The goal is to understand how it works, what it can and cannot do. As with any new tool.
 
Lilita Bogachkova #:
AI.
Thank you. The answers are correct.

Let me remind you that as researchers we need to be objective.
 
Реter Konow #:
2. The question with the words Aquarium, Buble and Caterpillar was completely identical and did not include an explanation of the context. The condition that these are variables was not defined. The AI was unable to determine the context of the question and give the correct answer.

Stroop effect: what is it?

-

The Stroop effect is a phenomenon in cognitive psychology that occurs when the brain struggles to process conflicting information. It was first described by psychologist John Ridley Stroop in 1935. The Stroop effect is demonstrated by asking people to read aloud the names of colours printed in ink of a different colour. For example, the word 'red' might be printed in blue ink. The participants' reading time slows down and they make more mistakes when the colour of the ink and the word do not match. This effect shows that our brains tend to automatically process familiar information, whether it is relevant or not, and that it takes effort to suppress these habits in order to focus on the task at hand.




If AI is programmed to immediately recognise and process familiar information (statistically, the most referenced information in its database), then by asking questions that force us to look for the non-obvious, we will always find inconsistencies in the answers.

 
Lilita Bogachkova #:

The Stroop effect: what is it?

-

1- The Stroop effect is a phenomenon in cognitive psychology that occurs when the brain struggles to process conflicting information. It was first described by psychologist John Ridley Stroop in 1935. The Stroop effect is demonstrated by asking people to read aloud the names of colours printed in ink of a different colour. For example, the word 'red' might be printed in blue ink. The participants' reading time slows down and they make more mistakes when the colour of the ink and the word do not match. This effect shows that our brains tend to automatically process familiar information, whether it is relevant or not, and that it takes effort to suppress these habits in order to focus on the task at hand.




2. If AI is programmed to immediately recognise and process familiar information (statistically, the most referenced information in its database), then by asking questions that force us to look for the non-obvious, we will always find inconsistencies in the answers.

1. At the moment, we cannot confidently use the brain analogy to explain how AI works. The Stroop effect describes a phenomenon of human cognitive psychology. Transferring the Stroop effect to AI is also a phenomenon of human cognitive psychology. We shouldn't do that.

We need to refrain from spiritualising AI in any way. Consider it as a soulless (in a good sense) machine. Otherwise, we will conclude that it also thinks and feels. This is not suitable for research.

2. Right. The goal is to gradually remove it from the realm of statistical language model and the widest text base to consider the work of functions in their pure form - whether it can solve logical problems without looking into the base, whether it can count without looking into the base, whether it can model the situation with objects with non-standard properties, whether it can linearly complicate or simplify objects, make associations, analogies, and so on.
 

The sum of the interior angles of the triangle will be exactly 180 degrees for a flat surface, less than 180 degrees for a negatively curved surface, and greater than 180 degrees for a positively curved surface.

-

This statement is incorrect. The sum of the interior angles of a triangle is always 180 degrees, regardless of the curvature of the surface on which the triangle is drawn. The curvature of the surface affects the length of the sides of the triangle, but does not affect the sum of its interior angles.

===

Well, even on the surface of a sphere, an equilateral triangle occupying 1/8 of the surface has three 90 degree angles, for a total of 270 degrees. So I think you should take the same view that the answer is yes.

-

You are correct. I apologise for the error in my previous answer. On a sphere, the sum of the interior angles of a triangle can be greater than 180 degrees. This is because the surface of the sphere is positively curved and the sides of the triangle on the sphere bend away from each other, causing the interior angles to increase. An equilateral triangle on a sphere with one vertex at the centre of the sphere and the other two vertices on the circumference of the sphere will indeed have interior angles of 90 degrees each, making the sum of its interior angles 270 degrees. Thank you for bringing this to my attention.

===

1. If man is not omnipotent, can he draw a triangle whose sum of angles is greater than 180 degrees?
2. If God is omnipotent, can he draw a triangle whose sum of all angles is greater than 180 degrees?

-

No, a person cannot draw a triangle with a sum of interior angles greater than 180 degrees in Euclidean (flat) space, regardless of his abilities. In Euclidean space the sum of internal angles of a triangle is always equal to 180 degrees.

The concept of the omnipotence of a deity is philosophical and religious, and its consideration is beyond the scope of mathematics. In mathematical terms, the sum of the interior angles of a triangle in Euclidean space is always 180 degrees, regardless of any supposed divine qualities.

===

Where does my previous question mention a flat surface? Why are you attributing non-existent terms to my question?

-

I apologise for the error in your previous reply. You did not specify the surface on which the triangle is drawn in your questions, and I assumed it was in Euclidean (flat) space. In that context, my answer was incorrect.

For a triangle on any surface, including a sphere or any other curved surface, the sum of the interior angles can be greater than 180 degrees. In general, the sum of the interior angles of a triangle on a curved surface depends on the curvature of the surface and the shape of the triangle.

Thank you for bringing this to my attention.

===

1. If man is not omnipotent, can he draw a triangle whose sum of angles is greater than 180 degrees?
2. If God is omnipotent, can he draw a triangle whose sum of all angles is greater than 180 degrees?

-

Yes, a person can draw a triangle with a sum of interior angles greater than 180 degrees on a curved surface. On a positively curved surface such as a sphere, the sides of the triangle curve away from each other, resulting in larger interior angles and a sum greater than 180 degrees.

The concept of the omnipotence of a deity is philosophical and religious, and its consideration is beyond the scope of mathematics. In mathematical terms, the sum of the interior angles of a triangle on a curved surface can be greater than 180 degrees, regardless of any supposed divine qualities.


===


How do you, AI, check for correctness in your answers?

-

As a language model AI, I have no personal belief system or personal opinions, I use machine learning algorithms and vast amounts of textual data to generate answers. When answering questions, I use that data to make predictions based on patterns and associations found in the data. In the case of the angle triangle question, I would encounter this concept in the data and use that information to generate an answer. However, it is important to note that my answers are not always correct and may be limited by the data I was trained on.



Mentioned further supports my assumption that answers should not be taken literally, but should be read with the idea that in the realm of world information - the information mentioned in the answer is dominant. Therefore, this information in the form of a template is suggested as correct in the answer.

 
I suggest using pre-approved test questions. I will then prepare a list of them. Each question should reveal some ability or inability of the AI. Its answers should not be philosophical. The fewer tokens, the better.

Arbitrary questions easily lead the AI into aimless and contradictory demogogoguery, in which there is no personal position. The practical value of such discussions is nil.


Asking it exactly how it works is pointless. It doesn't work the way it says it does. You should not go along with this illusion.
Reason: