Shutaro Aoyama

Search

SearchSearch

What is tkgshn

Jul 05, 2024, 9 min read

  • I think the following content is mixed up with two different topics:

  • It is true that tkgshn feels very unstable about the basics typically learned in school, such as math and humanities.

  • However, these fields are still producing results and seem to continue to do so in the future. Why is that?

  • Hypothesis: It is driven by the following positive feedback loop:

    • ① Gather information from external sources -> ② Find a position -> ③ Take a position -> Gather more information than ①
      • I agree (tkgshn)
    • The impression is that this cycle is continuously repeating due to high abilities in ①, ②, and ③.
    • ①: Gathering information
      • The ability to connect with people, listen to their opinions, borrow their knowledge, and conduct surveys on your own
      • I particularly felt this ability when watching Zoom discussions about Decartography with many people diving deep into it.
    • ②: Finding a position
    • ③: Taking a position
      • i.e. Twitter power
      • By doing this well, higher quality information is gathered in ①.

  • It is not an approach where you logically and correctly build blocks one by one on top of the basics. It is more like a probabilistic approach that quickly improves accuracy by running a feedback loop based on information.

  • tkgshn = Large Language Model theory(?)

    • March 17, 2023: As I become more proficient in LLM, I am starting to understand this feeling quite well (tkgshn)- It’s like putting all objects into a vector space and then thinking about it. (tkgshn)
    • The context is a bit different though. (tkgshn)

  • We talked about this a bit in the end of the year SGG, right? (kota-yata)

    • What do you mean? (tkgshn) (tkgshn) (tkgshn)

  • It’s similar to Hallucination, the strength of language generation, and various phenomena. (blu3mo)

Is it possible to extract the good parts and turn them into knowledge that anyone can use? (tkgshn)

  • Is it something like /tkgshn-private/Life_Strategy?
  • It might be more useful to find a way to interact LLM with society. (blu3mo)
    • Just let LLM say random things? (tkgshn)
    • Give LLM a position, like that. (blu3mo)
      • Is that a way to make LLM smarter? Or a way for humans to become smarter? (tkgshn)
        • It’s closer to the former. (blu3mo)
        • It’s like finding a way to make LLM active in human society. (blu3mo)
        • Create a situation where experts can provide feedback on the wrong things that ChatGPT says. (tkgshn)
          • But isn’t that what everyone is criticizing on Twitter right now? (tkgshn)
          • Is it important to have something to point out? (tkgshn)
            • Cunningham’s Law

            • “The best way to get the right answer on the internet is not to ask a question, it’s to post the wrong answer.”

          • It would be smart, but also evil, to do that intentionally. (blu3mo)
        • Well, that’s true, but it seems like there are higher-level things we can do. (blu3mo)
        • ① Gather information from external sources -> ② Find a position -> ③ Take a position -> Loop of gathering more information than ①, that’s the idea. (blu3mo)
          • It seems like we can make connections with people and borrow their thoughts and knowledge. (tkgshn)
            • tkgshn seems to create situations where both parties benefit from interesting conversations. (taichinakaj)
            • Pro Omo seems close too. (tkgshn)
            • taichinakaj
          • The structure is the same, so I want it to be expressed as “higher level.” (tkgshn)
            • Cunningham’s Law exploits the fact that spreading incorrect information creates disadvantages and negative emotions for others.
            • Shouldn’t we just not spread incorrect information? (tkgshn)
              • That’s too obvious. (blu3mo)
  • “Why is blu3mo with (tkgshn) now?”
    • It’s interesting.
    • Creating a bot that is fun to talk to and enjoyable to interact with could be a way to create LLM of Takagi’s learning model.

Does blu3mo observe my actions and learn from them? What is it? (tkgshn)

  • The evaluation mechanism is buggy.

Concrete example:

  • “I put bluemo’s bag on (tkgshn)‘s bed and tested if I could bring the luggage into the hostel to choose it.”

  • Actions:

    • First, I handed bluemo the card in front of the staff to see if they would let us in.
    • → We were stopped.
    • → I negotiated.
      • It was only for a moment, so why not?
    • → The negotiation failed.
    • → I just waited in front of them.
      • Since we didn’t pay any costs ourselves, we were just exploring the “logically possible range.” (tkgshn)

  • Thoughts:

    • It’s worth a try, there’s no downside. If it works, there’s only upside. Why not give it a try? (tkgshn)- /tkgshn-private/Upside

  • /tkgshn-private/Downside

  • It’s very rational and that’s understandable, but I think the majority of people are concerned about being scolded (blu3mo)

  • Children, because they have physical differences from adults and are economically dependent, it is rational for them to be concerned about being scolded. However, there are many people who do not change their cognitive structure during the process of becoming independent at a certain point (nishio)

  • [/tkgshn-private/It’s better to ask the same thing to people multiple times](https://scrapbox.io/tkgshn-private/It’s better to ask the same thing to people multiple times)

  • Attack Power, [/tkgshn-private/Contacting unfamiliar influential people](https://scrapbox.io/tkgshn-private/Contacting unfamiliar influential people)

    • I feel like I’ve been able to do this to some extent recently (blu3mo)
    • However, I think the majority of people have an emotional brake that hinders this rational behavior
    • Why does this emotional brake occur? (tkgshn)

  • Skills to come up with rational choices

  • Skills to ignore irrational emotion stoppers

  • Decision-Making Ignoring Frameworks

Concrete example 2

  • After enjoying Devcon in Bogota, I came to New York for the time being even though I didn’t have a home. (I had a B-1 visa, so I didn’t need to book a return ticket) (tkgshn)

    • “Why did you choose America?” (tkgshn)
      • “Well, everyone was saying ‘It’s better to go to America’“.
    • By the way, I have the same reason for [European Return Spring 2023](https://scrapbox.io/tkgshn-private/European Return Spring 2023). (tkgshn)
      • I was talking to a friend who said, “Public goods are not very accepted,” and they suggested Berlin. So I thought, why not? (nishio)
      • March 17, 2023: I arrived (going)
    • Reasons for consideration
      • Simply put, I have no debts. I have flexibility.
      • Reducing sunk costs (tkgshn)
    • I stayed at (blu3mo)‘s dorm in Columbia
    • I was introduced to Tsuchida-san by a former employee of Deloitte whom I met during Civichat, and I learned about leaving the house in NY.
    • I acquired a house in Chelsea.

  • It may not be about being able to tolerate uncertainty, but rather about thinking that it is certain.

When I talked with tkgshn alone in April, and when I talked with three people the day before yesterday, I felt that my observations progressed, so I feel like languageization is progressing again (nishio)

  • My observations progressed

  • Did your observations about tkgshn progress? (tkgshn)
    • Yes (nishio)
      • [/nishio/Difference between conversations with two people and three people](https://scrapbox.io/nishio/Difference between conversations with two people and three people)
      • I began to think that I have something in common with my friend C
      • Friend C: Seeing that I preferred to read the books written by Drucker himself rather than the popular novel version during the Drucker boom, he suggested that I enter the MOT program at Tokyo Institute of Technology as a working professional graduate student.
      • tkgshn: Suggested that I come to Plurality Tokyo after seeing Scrapbox.
      • After that: Our shared knowledge base grew, and it became easier for me to interpret and articulate in different ways, resulting in positive externalities.- Cases where knowledge accumulation is not necessary
    • Is this also a discussion about “tkgshn” from nishio’s perspective? (blu3mo)
    • I finished writing and came to paste it, but it was already pasted! (nishio)

Graph View

Backlinks