AI 2023. Meet ChatGPT. - page 169

 
I should add: not only select and combine sequences, - Transformer can create similar data sequences on the basis of statistical model. And that is generation.

What strikes me most of all, in modern AI, is a phenomenon called "emerging properties". It seems to be something magical, although it can be explained technically. Still, I cannot yet fully grasp it with my mind.
 
Dmitry Fedoseev #:
Because they're not doing well... That's why it's this "going to the people" thing. When they do, they'll close up shop.

It's possible. But the sources won't go anywhere, the Internet remembers everything, even if someone wants to delete it, it will appear on torrents. If a free and open source product becomes cooler than a paid one, the company will get nothing. And right now, they're taking a huge loss. All so that everyone can have AI....

Retag Konow #:
2. I think this shows a deliberate devaluation of AI. In fact, democratisation of AI (as they call it in western publics) means competition between companies.

On the other hand, the provocateurs of "democratisation" may be providers-sellers of computing resources, such as Nvidia, Meta and others.

Competition will lead to free products being so cool that the benefits of paid products become negligible.

But the idea that hardware companies will benefit looks interesting.

Crypto has already shook the hardware market, but with the regular improvement of AI it can be done for a long time. Computing power will always be scarce.

Maybe because of AI computers will start to become obsolete again and it will be necessary to buy new DDR7-8-9 etc., which will revitalise the computer market.


Next is my fantasy, which is based on some facts.

I remember there was information about the update of Windows from the spectrum virus and something else, that in the end such an update led to a decrease in processor performance by 30 per cent. Then the processors were already released with a hardware system that reduced performance.

Even before this update, many computers were infected with hidden miners, which also led to performance degradation. But that's software, not something hardware.

I can't say it's impossible, but the likelihood of your PC being used for general computing needs by someone, is real.

Since AI consumes overdrive, getting people to upgrade their PCs (due to the hype of free AI available) so that the shared network can then reduce costs for some large corporations is also realistic.

 

Found this on the internet.

3 promts that will squeeze the most out of ChatGPT. Enthusiasts have already compared their effectiveness from a scientific point of view - with them you will really get more accurate answers.

The video is in English, but you can watch it through Yandex.Browser in Russian.


Universal promt for any query for a more accurate answer. Neuronka breaks questions into steps and answers each one in order. It is easier to identify errors and correct the query.

Let's work this out in a step by step way to be sure we have the right answer.

Перевод: Давайте разберемся с этим шаг за шагом, чтобы быть уверенными, что у нас есть правильный ответ.

You don't need to google to double-check ChatPGT - use this promt. Neuronka will analyse its own answer and highlight weaknesses in detail.

You are a researcher tasked with investigating the *что проверить* response options provided. List the flaws and faulty logic of each answer option. Let's work this out in a step by step way to be sure we have all the errors:

Translation: you are a researcher tasked with examining the provided answer choices *what to check*. List the weaknesses and faulty logic of each answer choice. Let's break it down step by step to make sure we have all the errors:

Edits from the past promt can also be assigned to neuronkey itself. To do this, create a separate "personality" with this promt.

You are a resolver tasked with 1) finding which of the *какие правки* answer options the researcher thought was best 2) improving that answer, and 3) Printing the improved answer in full. Let's work this out in a step by step way to be sure we have the right answer:

Translation: you are a solver tasked with 1) determining which of *which corrections* the researcher thought was the best answer, 2) improving that answer, and 3) printing the improved answer in full. Let's break it down step by step to make sure we have the right answer:

 

Promt GPT for vendors:

Рассмотри возможные возражения против [продукта/услуги] и дай пошаговую инструкцию, как ответить на эти возражения таким образом, чтобы [продукт/услуга] понравился покупателям.


 

Employment Prompts


 

I have just been shown another miracle grail on the main page of the forum with the magic letters GPT in its name. For just a tonne and a bit of baht you will be connected to the magical world of AI and you will start earning ******/day. Naturally, lying on a sun lounger in the Caribbean. Surrounded by sultry beauties. Without doing shit! ))

From my experience, so far GPT is only good for writing separate functions or methods of a class on a very clear assignment. It can quickly configure a dummy class. But for that I need to write a list of empty variables and methods for it. And the writing will be much more than filling it all in by hand in the editor. But surely there will be a bunch of suckers who have read the internet pop about the magic features of GPT and will buy this scam.

 
Alexey Volchanskiy methods of a class according to an absolutely clear task. It can quickly configure a dummy class. But for that I need to write a list of empty variables and methods for it. And the writing will be much more than filling it all in by hand in the editor. But surely there will be a bunch of suckers, who have read the internet-pop about the magic features of GPT, who will buy this scam.
Looks like you've turned 180 degrees from your previous stance on GPT. )) Why all of a sudden, if it's not a secret?)))
 
Реter Konow #:
Looks like you've turned 180 degrees from your previous position to GPT. )) Why all of a sudden, if it's not a secret?)))

Mmmm, what was my position before and what is it now, if you look from the outside, for example, through your eyes? )) My internal position since the beginning of my acquaintance with GPT - exploring the possibilities of the technology and how I can make money from it?

I have never been a fanatic of any idea and always tried to be realistic.

For example, in my neighbouring chain store, where I buy groceries, there are only Muslims working there. And according to my tradition, I got to know most of the girls, I know their full names (often Russian names or abbreviations are on their nametags), who they are from, etc. Recently in spring was the month of Ramadan, they are not allowed to eat and drink from sunrise to sunset according to their religion. I asked ChatGPT to write encouraging greetings during this difficult period. I am zero amp in Islam, and he wrote a very beautiful congratulations in oriental style and even inserted a quote from the Koran in Tajik at the end - I have a favourite (platonically) Tajik girl there. Just in case, I asked him to translate it into Russian, there was a good quote, something about perseverance and faith. I printed it out and gave congratulations to all the girls.

And taking into account that I told the Tajik girl that as a student I read the medieval Koran in the Public Library (I was studying world religions), I am now enrolled in the category of sympathisers. Well, remember, in 1917 there were Bolsheviks and sympathisers :))).

I should try to connect to the 4th version, although it is impossible to do it directly from Russia under sanctions. They say that GPT 5 will be released at the end of the year, so it won't be boring :).

 
How to bypass Anti-Plagiarism so that it does not scorch ChatGPT.

Shortly about the most important thing: as soon as Anti-Plagiarism clings to the text - return it to ChatGPT and use the magic Promt. Uniqueness jumped from 0.07% to 94.89%.

Magic Promt: Rewrite the text, change the structure and make it more readable. It must be original, avoid repetition and pass the Anti-Plagiarism check: [insert text].

And a couple of tips on what ChatGPT most often falls foul of:

- There is a summarising part, a colon, then a qualifier
- Phrases in the text are repeated from paragraph to paragraph
- There are many nouns in a sentence in a row, reminiscent of text from old textbooks.
 

It's a sad outcome


Reason: