Sunset programming? - page 5

 
Andrey Pogoreltsev:

Don't confuse programming and software development. Programmers died out a long time ago. The developer army is only growing, because it's not silly clicking the keys, throwing shapes and patterns on a GUI.

And it's always business as usual for cheap resources. An AI creating a calculator will cost significantly more than the resources of the developer of that calculator, so the author's original question is more hypothetical than real.


The complexity of the products also varies greatly. For example, until someone creates an AI that can create something like ZCach with zkSNARKs, it will take many years to come up with even more complex algorithms.

And tell me how it is different?

 
Dmitry Fedoseev:

And tell me how it is different?

Programming is just one stage of development.

 
Реter Konow:

There are often headlines predicting the end of the programming era. They say that AI and neural networks won't stand a chance against humans. I used to mockingly read such articles, but now I am not laughing. After all, Objects (which OOP creates) consist of parameters, like bodies of particles, and are connected by links. Parameters and links do not have to be described in code, especially - in different languages. They can be assembled in the studio with a set of visual tools and any program can be written that way. I wonder why development has not gone this way...

AI does not equal intelligence, nor does dumbcoding equal programming.

Just yesterday I was reading an article almost on the subject on hubra: https://habr.com/ru/post/483592/

Разработчики — никакая не элита, а голые короли индустрии
Разработчики — никакая не элита, а голые короли индустрии
  • habr.com
Пожалуйста, не поймите меня неправильно. Я профессиональный разработчик с 30-летним стажем. Я могу читать и понимать почти двадцать языков, полных по Тюрингу. Я могу писать ясный и выразительный код на доброй дюжине из них. В языках, которые я действительно люблю и использую ежедневно — мое имя можно увидеть на доске почета StackOverflow...
 

Well, sooner or later the sunset will come. True, it will most likely look like a major programming discontinuity. Even more high-level languages with most likely choice of direction of neural networks and their interaction. But that will be programming too. And there will also be those who do the layout of websites through studios. True, the difference between these programmers will start to grow a lot.

And full decline will come when 90% of all businesses will be fully automated, as well as medicine and other ... that is, utopianly not yet a long time from now.

 
AI (conditional) wins in computer games of 3d-shooters against champions(nerds), where both are put in the same conditions and most importantly, it is not a static calculation of the move, but a dynamic one. I.e., it's not a turn-based game (checkers, chess), but an element of action where victory depends not only on intelligence, but on the player's reaction. I.e., weapons have recoil, the player (person) came around the corner, he can both go around the other side, can sit waiting, can run into a distant room. And in such circumstances, the AI has to think-over-educate (learn) to "run successfully, shoot accurately".

It's just a minor detail, now this AI is being implemented everywhere, and all efforts are thrown into its development. So the next revolution is AI
 
Alexandr Andreev:

Well, sooner or later the sunset will come. True, it will most likely look like a major programming discontinuity. Even more high-level languages with most likely choice of direction of neural networks and their interaction. But that will be programming too. And there will also be those who do the layout of websites through studios. True, the difference between these programmers will start to grow a lot.

And full decline will come when 90% of all businesses will be fully automated, as well as medicine and other ... which is a utopian time.

A programming language in itself is inefficient in principle. Rules, syntax, typing of pages and code reading. If we perceived the world of real objects through their code, we would look at it through the eyes of amoebas.

All necessary frameworks have been created long time ago and modern programmers just bolt on "spare parts" from various Objects. This, by the way, is often written about in various topical articles. The development of programs has passed into upgrading them. New ones are rarely created. Mostly variations of the old one.

The essence of my idea is that we are using too narrow and laborious method of describing program objects. Modern technology allows us to expand it and improve it many times over, but for this, we need to reconsider the concept of programming.

 
SeriousRacoon:

AI does not equal intelligence, nor does dumbcoding equal programming.

Just yesterday I was reading an article almost on the subject on hubra: https://habr.com/ru/post/483592/

That's right, not intelligence. AI is a very broad field and refers to literally anything that uses fuzzy logic (which is its similarity to us).
 
Ivan Butko:
AI (conditional) wins in computer games of 3d-shooters against champions(nerds), where both are put in the same conditions and most importantly, it is not a static calculation of the move, but a dynamic one. That is, it is not a turn-based game (checkers, chess), but an element of action, where victory depends not only on intelligence, but on the player's reaction. I.e., weapons have recoil, the player (person) came around the corner, he can both go around the other side, can sit waiting, can run into a distant room. And in such circumstances, the AI has to think-over-educate (learn) to "run successfully, shoot accurately".

It's just a minor detail, now this AI is being implemented everywhere, and all efforts are thrown into its development. So the next revolution is AI
Agreed. But, you can't create AI on neural networks alone. You need a concept of intelligence, and it does not exist in its ready-made form yet. Only theories.
 

A method to create a visual studio from absolute zero in a pure EA, using only OnChartEvent() events - coordinates and cursor and key events. According to my theory - it's possible. I use my new concept of assembling object parameters.

Stages:

1. Create a new EA.

2. Declare Kernel array.

3. Write a function to control the size of the kernel with the cursor when the key is pressed.

4. Write the function of adding parameter to the core by pressing the key button.

5. Add calls of basic handlers of graph objects on key events of cursor and special button (handlers - ObjectCreate(), ObjectSetInteger()...).

6. Write a function to save the object's template.

7. Writing function that copies the template into the instance.

8. Write a function that distributes the created templates and instances within the kernel.

9. Run the Expert Advisor and proceed to creating the parameters and assembling the templates of basic controls on the graph of MT.

10. Having created the basic set of elements, we save them as templates and copy them into the user part of the kernel. From the templates of the first elements we get instances of elements for editing.

11. Establish links between the parameters of the editing and editable elements and get visual editing)). (albeit in a primitive and crude form, but that's essentially it).


ZS. As you can see, you can get by with almost entirely visual assembly of controls and implementation of visual editing.

Документация по MQL5: Основы языка / Типы данных / Структуры, классы и интерфейсы
Документация по MQL5: Основы языка / Типы данных / Структуры, классы и интерфейсы
  • www.mql5.com
Структура является набором элементов произвольного типа (кроме типа void). Таким образом, структура объединяет логически связанные данные разных типов. Объявление структуры Имя структуры нельзя использовать в качестве идентификатора (имени переменной или функции). Следует иметь ввиду, что в MQL5 элементы структуры следуют непосредственно друг...
 
Andrey Pogoreltsev:

Programming is just one stage of development.

What others are there?