AI 2023. Meet ChatGPT. - page 104

New comment
 
Grigori.S.B #:

ChatGPT AI Made Me A $100,000 TRADING STRATEGY

https://www.youtube.com/watch?v=Jh5rJskkEkU& ab_channel=HumbledTrader

Moving average trading on nasdaq stock, yes, the grail)

 
Vitaliy Kuznetsov #:

Moving average trading on nasdaq stock, yes, the grail)

Also such a screaming round sum

 

Why develop special "thinking mechanisms" if a statistical model based on textual data reproduces "thinking" mathematically?

Everyone knows that every sentence and every word of text data contains human thoughts, but not everyone knows that the statistical approach extracts and"weighs" chains of text representing thoughts (knowledge and attitudes of people) in order to "copy" interrelations and generate new chains (thoughts) according to mathematically verified patterns.

Any thought represented in a text is transformable into a chain of tokens. "Weighing" the chain on the basis of textual material, unites it by connections with the context group (chains or thoughts, in human understanding). That is, each chain has its own context space consisting of links to other chains. The common context space of all chains is the space of AI "thinking".

Why, in this case, develop the technology of "real", knowledge, feelings, relationships? Why not "copy" the products of human thinking from the text and claim the "reasonableness of AI"?

And what is "Reasonableness" in this context, as not a well-working and well-adjusted statistical model, which determines the best chances of survival and prosperity in the surrounding world?

 
There is no "AI." There is a statistical approach by which people "copy" human thinking from texts, and by additional algorithms complete the illusion.
 
Реter Konow #:

Why develop special "thinking mechanisms" when a statistical model based on textual data, reproduces "thinking" mathematically?

...

To date, short episodes of the "thinking" process are reproduced as part of the response to input prompts. Although it is clear that there is no "thinking" involved. At least not in the human sense.

Human text represents the world only to humans, but to a machine, the text represents the next, context-bound text. Or other data.

Let's imagine that we loop the model in the process of "thinking" by making it set itself tasks and solve them. There are already such attempts (Auto-GPT, babyagi). What is the use? Well, the model will become uselessly generating content nobody needs. In fact, in a modified form, it will reproduce the original textual content it was trained on. Nothing new will come out of the recombinations.

But how can we make it work?


 
Perhaps we need the text to represent the world to the machine as it does to the human. What's the point of text, though? We need to train the model on real world data. But I don't think that's enough power.
 

He's still confused :)

ChatGPT programming

Told him he was wrong ) He corrected himself ))

ChatGPT is programming in mql5

 
Реter Konow #:
To date, short episodes of the "thinking" process are reproduced as part of the response to input prompts. Although it is clear that there is no "thinking" involved. At least not in the human sense.

Human text represents the world only to humans, but to a machine, the text represents the next, context-bound text. Or other data.

Let's imagine that we loop the model in the process of "thinking" by making it set itself tasks and solve them. There are already such attempts (Auto-GPT, babyagi). What is the use? Well, the model will become uselessly generating content nobody needs. In fact, in a modified form, it will reproduce the original textual content it was trained on. Nothing new will come out of the recombinations.

But how can we make it work?


How will you take the information that there is a chemical soup inside your head that has imagined what it thinks? ))

 

He's also smart ))

But he got better again ))

ChatGPT is programming on Mt5

 
Alexandr Bryzgalov #:

He's also smart ))

But he got better again ))


ask the same code in MQL6 ;-)

1...979899100101102103104105106107108109110111...214
New comment
Reason: