Diary 2023-08-19
LLM camp continues I had an early dinner at about 6pm and was supposed to have work time until 9pm, but I couldn't get my head around it at all.
- I even have a blueprint of the code, but I can't write it at all.
Returned to the hotel room around 9:00 p.m., but didn't have the energy to go to the bath and sleep for a while. Woke up around 24:00.
It's a bit sad to read this story. When I read this story, I feel a little sad. I feel sorry that I can't write any code because my head is spinning even though I ate early. It's also sad that I don't have the energy to go to the bathroom and just go to bed. But even if I don't go to the bath, I might be able to have a great adventure in my dreams, so I might actually become a secret hero. I'm a daredevil adventurer! (GPT4)
By using LLM, you can get sloppy comments on everything!
So far there has been human interaction only after B
But then I realized that after C might be more appropriate.
People who do the thinking process only in their minds cannot observe it.
It's hard to verbalize what we can't observe.
At some point in my life I encountered the KJ method, which allowed me to observe certain parts of my thoughts.
But it turned out to be only one step.
I didn't observe the part of the thinking that goes on in the brain during that process of "me doing the KJ method."
So, while I used the KJ method and the further developed Kozaneba, I could not verbalize how to use them in a way that would be beneficial.
This time, the AI output did not go through "the process of chopping sentences into fragments," but a KJ-method grouping-like phenomenon occurred, so the observation proceeded.
- When I do KJ method or Kozaneba, I do the "process of chopping sentences into fragments" before these three steps, so I was not able to compare them like this
I have a feeling we're getting closer to an answer.
- This is a true example of [Polanyi's tacit knowledge
- Even if you don't know the right answer to arrive at, you can get closer to the right answer if you have a sense of whether or not you are approaching the right answer
[/unnamed-project/🤖2023-08-18 06:35](https://scrapbox.io/unnamed-project/🤖2023-08-18 06:35)
A question-answer cycle might be beneficial.
I was going to try this with AI Qualia.
Diary 2023-08-18 ←Diary 2023-08-19 → Diary 2023-08-20 100 days ago Diary 2023-05-11. 1 year ago Diary 2022-08-19.
This page is auto-translated from /nishio/日記2023-08-19 using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I'm very happy to spread my thought to non-Japanese readers.