BELOW_IS_LESS_INTERESTING
Digest of Research Notes
.
We are exploring a method of enumerating the contrast between two metaphors on GPT4 in the space of metaphors, abstracting them, and applying them to different domains to make them concrete. This is related to symbolic modeling and equivalent transformation theory.
We had GPT4 enumerate the differences between the worldviews of nomadic and agrarian peoples, abstracted them, and concretized them by applying them to the way we work in developed countries in the age of AI. Specifically, we used the following contrasts: fluidity vs. sedentariness, nature symbiosis vs. nature domination, communalism vs. individual ownership, flexible vs. fixed social structures, nature faith vs. agrarian faith, oral vs. sedentary culture, nomadic war vs. territorial defense war.
In addition, βnatureβ is a metaphor for the environment and conditions in the working style and organizational culture. And we believe it is important to view βescapeβ in terms of situational judgment, risk management, flexibility, learning opportunities, and self-protection.
Considerations and questions
This approach would be useful for transferring ideas between different domains and providing new perspectives. However, how much information does GPT4 require for proper abstraction? And what level of domain knowledge is needed for embodiment? These questions will be clarified in future research.
extra info
TITLES: Divergence and abstraction in GPT4 in metaphor space before concretization generated: 2023-08-18 11:55
previous notes
This page is auto-translated from /nishio/π€γ‘γΏγγ‘γΌη©Ίιγ§GPT4γ«ηΊζ£γ¨ζ½θ±‘εγγγγ¦γγε ·δ½εγγ 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. Iβm very happy to spread my thought to non-Japanese readers.