AI 2023. Meet ChatGPT. - page 99

 
Реter Konow #:

So, a person will voluntarily deprive himself of his body and become an iron (or not an iron, maybe plastic) with electronic stuffing of indeterminate geometric shape?


Of course voluntarily, of course)))) there are already examples.
and about energy you don't need to worry, the liquefaction of proteins and carbohydrates into water and carbon dioxide is an efficient but very primitive source of energy. energy surrounds us everywhere, new organisms will not need to eat something or someone all the time.
 
Andrey Dik #:

of course voluntarily, of course))) there are already examples.

Yes, I did not take into account such a variant of the historical process)).

 
Реter Konow #:

Yes, this variant of the development of the historical process I did not take into account.))

Well, you see!)))) anything can happen, up to the unthinkable.

there is always a certain chain of events, the universe is so organised, you can't confuse the effect with the cause, although it's not exact.

perhaps the quantum nature of the universe generates local time loops and probabilistic civilisations always create themselves. they return to the past and stir up protobullion, from which they themselves emerge later on.

Flying clots of energy - ball lightnings, behaving very sensibly at times, are not mechanical civilisations? Flying saucers and other "alien stuff" are not local species of beings from the future?

all over the world they find ancient ruined, melted down radioactive cities. creatures from the future think "whatever we do, these ancient sheep end up in a nuclear disaster!" .....

 
 

Well, I decided to look for a practical application for GPT

It's pretty good at creating chords for songs. Whether it finds them in the database, I don't know.

I also liked the promt feature, "Write 10 shocking facts about ________________."

Sitting there, dabbling for an hour, probably. In general, lists (various) rule.


Learned a new word "Overshocking", which does not know Yandex and Google).

Turned on logic and tried the translator.



I think the main practical side of GPT is recycling information (shortening, squeezing out the main thing). By the way, there are 2 companies that do business on selling "Book Summaries". I think GPT will help them a lot.

I have one folder of docs that I have been collecting for a very long time.

As soon as the technology becomes available, I'll pop it into a bot and ask it to write a book that contains a logical sequential narrative that contains a squeeze of at least 80% of the material provided, and 20% it can take from its base for a logical connection.

 
They say GPT 5 will be indistinguishable from a human being.

I hope they'll let me download it to my computer. I want a mate, a wife and a lover, and I don't want my secrets to be sent to the internet. I'll set it up for myself and communicate with all three of them in the allotted time.

And in general, it's time to make a real personal account for the chat room. You pay for tokens/memory. Chat adapts to you, remembers all your correspondence with him, has access to your phone, calls you and talks to you in a voice you have set up to remind you about Gali's birthday, because you forgot last time it was ugly.

I predict that LC with a personal chat. Soon. Another +$20 subscription +$10 per memory unit.
 

Want to know how strong GPT-4 is in programming?

Watch the video: https: //www.youtube.com/watch?v=IyKKhxYJ4U4&list=PLwsc0Oqqkbu0fJrPWLFf9heFxzJ4R6t4N&index=1


1- A blogger, who doesn't know programming and is not oriented in the development environment, sets himself a task, to write a simple game on java script, using GPT-4 as his personal freelance programmer. From his own skills, only on copypaste. Does not read or understand the code himself.

2. For code puts Visual Studio editor. For generating images (backgroud and character), turns to Midjorney.

3. TK is a simple Mario-style game. The character runs on platforms hanging in the air, jumps over gaps, collects coins. 2D space. The simplest physics. Movements: forward, backward, jump. If falls down from the platform - Game Over.

4. Additional tasks in the TOR:

  • Animation of leg movement when walking.
  • Picture of the character exploding when falling down into the lava.
  • Scrolling background as the character moves forward.
  • Game restart function and button.
  • A function and points window for counting coins collected.
  • Game can last indefinitely, until lost and restarted.

5. Blogger turns to GPT-4, puts out the ToR and asks to write the code of the game.

6. During many hours he has a dialogue with GPT-4, in the following format:

  • Asks for code that executes or changes a certain feature in the game.
  • Receives the code, inserts it into the editor, compiles it, evaluates the result, finds inconsistencies.
  • Goes back to GPT-4, states what didn't work or didn't work well and asks to redo or add something.
  • Gets new pieces of code and explanations of where to insert them. Copies it.
  • Goes back to VS, pastes, runs, looks at the result, goes back to GPT-4.
  • And so on and so forth for hours.

In the end, the blogger finally got the desired result from GPT-4, but you can evaluate what it cost him and what the result was only by watching the video.

I think this video should be watched by everyone who is afraid that soon they will start firing programmers. I've never seen a clearer demonstration of GPT-4's real capabilities in coding and, in general, in helping a person to solve a complex problem.

I Used AI To Make A Game (ZERO Coding Experience)
I Used AI To Make A Game (ZERO Coding Experience)
  • 2023.04.01
  • www.youtube.com
Here's a breakdown of how I used AI to make a video game.Here's the game: https://mreflow.com/jump-game/🛠️ Explore hundreds of AI Tools: https://futuretools...
 
Activating Windows 95 with ChatGPT
Activating Windows with ChatGPT
Activating Windows with ChatGPT
  • 2023.03.26
  • www.youtube.com
Hello, my friends! Let's hit 10000 likes? Check out my website! https://malwarewatch.orgToday I am going to activate Windows using ChatGPT. Ever since ChatGP...
 
Реter Konow #:

...

    5. Blogger turns to GPT-4, puts out the ToR and asks to write the game code.

    6. During many hours he has a dialogue with GPT-4, in the following format:

    • Asks for a code that executes or changes a certain chip in the game.
    • Receives the code, inserts it into the editor, compiles it, evaluates the result, finds inconsistencies.
    • Goes back to GPT-4, states what didn't work or didn't work well and asks to redo or add something.
    • Gets new pieces of code and explanations of where to insert them. Copies it.
    • Goes back to VS, pastes, runs, looks at the result, goes back to GPT-4.
    • And so it goes round and round for hours.
    ...

    It's worth noting here that at first the blogger didn't have a TOR, and didn't know how to make one. He asked ChatGPT, and the latter outlined a step-by-step plan for creating a game. Next, the blogger consistently entered the points of the plan into the prompt window, got pieces of code. Then, he copied, pasted, looked at the result and went back. And so on, with each amendment.

    Conclusions:

    • The AI could not write a game from scratch.
    • TheToR should be broken down into "digestible" parts beforehand.
    • Independently, consistently and persistently achieve the correct execution of each item, formulating what is wrong.
    • AI cannot visually analyse the result of its code in the context of the task.
    • A human has to act as a "fitness function" and manage the "optimisation" of the process: saying "good/bad" and how it should be.

    Suppose we decided to teach the AI to visually recognise a game to independently find errors, refactor the code and follow the gameplay specified in the TOR. Then, we will need video clips of dozens or hundreds of thematic games in conjunction with their code. How realistic this is, it's hard to say. There is an incredible amount of data. Learning from a stream of visual information is done by Tesla, but there is only one "gameplay" there, with a set of the same objects and rules.

     

    Any tool has to be smart to use. To me as a super smart manual is defined for today GPT. You don't need to read all the manuals, the main thing is to understand what should work and how and to ask the right question.

    SPSS IBM with the first time for half an hour understood how to make calculations of Spearman's correlation with GPT. I would probably read the manual for a day myself).

    The child's psychology diploma helped))))

    Reason: