🤖2023-08-19 07:08
What exactly is the source of discomfort in discomfort-driven intellectual production?
The "discomfort" in the probabilistic language model would be the low probability of a certain token being produced when conditioned on the surrounding token's situation (context).nishio.icon
Simply predicting the probability of token appearance in a small window size creates what appears to be "meaning" to humans ref: CBOW. But I don't believe that such a small context width is enough for intellectual production.
No, this is a difference in depth of intellectual production.
Is it the same thing to make a superficial expression natural and to make the meaning add up, or are they both the same?
Consistency in narrow context width = verbal fluency
Consistency across a wide range of contexts.
network of meanings
And how can we make use of it?
I think it's an important part in intellectual production.nishio.icon
In Kozaneba, the act of fragmenting sentences by removing particles, etc., and seeking a consistent spatial arrangement among them is an act of freeing oneself from "consistency in a narrow context width" and facilitating the search for "consistency in a wide context width".
Save bullet points in Scrapbox instead of "writing like a book or blog".
Linking between pages
These are writings that are more about "consistency across a wide range of contexts" than "consistency across a narrow range of contexts."
___BELOW_IS_LESS_INTERESTING___
Fragmentation of information and multifaceted associations create new meaning. This is accomplished by creating new connections to symbols that are not normally connected; it is even possible to export some data from groupware that the AI does not have direct access to. A deeper understanding of the process by which information fragmentation and multifaceted associations create new meaning is needed. The challenge for the future is how to integrate the specific implementation and functionality of "AI Dejima" and new perspectives on the process of information fragmentation and multifaceted associations into specific methods and tools.
In discomfort-driven intellectual production, a source of discomfort is necessary, and discomfort is unlikely to occur with the present self. It is not a reflexive response, but a slow storage of thoughts and then agitation. This is why we need to communicate with an AI that responds slowly, one piece a day, rather than a fast response.
Nishio's research notes confirm the importance of the source of discomfort in discomfort-driven intellectual production and the importance of storing thoughts carefully before stirring them. Also interesting are the perspectives of exchange diaries with AI and the vegetative growth of thoughts. Integrating these perspectives and incorporating them into concrete methods and tools will be a future challenge.
Question: What exactly are the sources of discomfort in discomfort-driven intellectual production? And how can it be harnessed?
extra info
json size: 41050194
pickle size: 778379445
previous notes size: 639
titles: erasing paths with eraser (is it necessary?) , Kozaneba Development Diary 2022-06-02, Dyson Sphere Program 2021-01-29, Hatena2008-10-24, Remote speaking can be recorded, Why is discomfort relevant when intellectual action is advanced with high load? , Japanese font, Opportunities for decision-making are a string of beads
---
This page is auto-translated from /nishio/🤖2023-08-19 07:08 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.