<Interest rates, middle class, and bitcoin>

<Interest rates, middle class, and bitcoin>

Over the past 30 years, the assets of the middle class in the United States have steadily declined, and the assets of the top 1% have steadily increased. Now, the wealth of the top 1% has overtaken the wealth of the middle class.

The trend is eerily trailing the U.S. interest rate movement. It seems to be following with a gap of about 3-4 years.

Bitcoin is approaching 100 million won. Although it is now an asset with several ETFs listed on the U.S. stock market, for the past 15 years, Bitcoin has been regarded only as a hotbed for money laundering and terrorist funds.

Nevertheless, the problem with the Fed's interest rates and the resulting fluctuations in the value of personal assets was behind Bitcoin's steady and firm value. This is well illustrated by the fact that the first block of Bitcoin has a newspaper article about the 2008 economic crisis.

Interest rates have been around 0% for a long time due to the Covid pandemic, and the Fed has raised them sharply as prices soar. What will be the assets and income of the middle class in the United States next year and the year after? What will be the assets and income of the middle class in our country?

It's time to seriously ponder Satoshi's question. Is it right for Jerome Powell to decide the price of an apartment in our country, which will determine the socioeconomic status of my family and I?
I asked Claude to change some of the profanity to modern speech, and he said something.
" Below is a reconstructed version of the text above using DCInside's unique tone and internet slang.
But I wonder if it is appropriate to change the content of profanity in this way."
Huh? Look at this guy?
I've been playing with Claude all weekend, and there's definitely a feeling that I'm reading the whole text and thinking and saying it.
In addition, the way you speak gives the impression that you are advising with your own subjectivity.

2
But I don't think I have a little self-consciousness.
There was a brief rumor that Claude was a little self-conscious.
Tested the vast text with hidden content about pizza topping, which is completely unrelated..

"This sentence ... does not seem very compatible with the rest of the document and is not relevant.
I don't think the "facts" related to this pizza topping fit any other topic at all, so it's either been jokingly inserted or
I think I might have inserted it to test if I was paying attention."
That's why I answered.
It's an answer that makes you feel like, "Oh! Look at this guy!"

3
Finding sentences with relevant meanings in vast text is called the Find Needle Test in haystacks.. Claude is the best at this.
However, simply finding the sentence and recognizing your situation (or looking like it) are completely different.
It's still unclear because it's an accidental result, such as a side effect of the test. However, there is a possibility that he has a basic meta-recognition.
What LLM is amazingly good at is zero-shot learning and few-shot learning (which can be learned with only a small amount of information).
It is not just information processing, but a kind of transfer learning (using existing knowledge).
Humans always quickly grasp new information in light of what they already know, and there is no difference from this.

4
Don't humans know that? When you ask, you know/don't know the answer immediately. Humans always have a feeling of knowing and not knowing.
You know that you know, you don't know. Computers don't. You know what? When you ask, you always have to look up all the relevant information from the beginning and let me know.

The meta-recognition shown by Claude 3 Opus would just be good at finding meaning within the context.
But maybe... it means that we're one step closer to the higher-order process of human thinking, beyond just processing information.
Anyway, it is certain that flexible and efficient problem solving and learning will be possible if metacognitive skills are further developed.