@nishio: In a world where huge language models (LLM) have been created and we can read more sentences than homo sapiens can in a lifetime and understand them better than homo sapiens, I’ve been wondering for the past few days what my raison d’etre is. I’ve been wondering for the past few days what my raison d’etre is in a world where people are reading more text than homo sapiens could ever read in a lifetime, and understanding it better than homo sapiens. I’ve come to a simple conclusion: “The ability to figure out what has not yet been written. - Not yet verbalized. / (6.2.5) That which has not yet been put into words
@nishio: In my life as a software engineer, there have been many times when I said, “I looked at the documentation and it doesn’t say that. I figured it out by experimenting with it myself. Science is the same, because what Homo sapiens doesn’t know yet has not been written, nor do we know the language model. Application of Knowledge is another.
@nishio: It is said in many fields that “doing what is written in a book as it is written in a book does not work” with respect to the application of knowledge. This is because books are the product of abstraction. When connecting to a concrete system, it is necessary to connect with the knowledge of the concrete system itself bouncing of ideas, opinions, etc. off each other to obtain a fine-tuned integrated whole. The book only talks about a few examples of specific systems - Books are already a product of abstraction - Performing stage of [U-curve model
@nishio: if you don’t understand the “system in front of you”, which is not written in the documentation, you cannot apply your knowledge to it. Whether software systems or human systems, most systems have only inadequate documentation. Understanding this cannot be assisted by a language model.
@nishio: i.e., the giant language model is a new expressway, a zone of “what has already been written” and takes us to the foremost line where we are confronted with “what has not yet been written” at high speed through the It is like a forest. There, we work to open up the forest of what has already been written. - Parable of the Forest : Thinking at the edge of the edge
@nishio: There are several variations of how to clear the forest, discovering new scientific facts is one of them. Creating implementations that work and are used in society is one of them, and observing and verbalizing what has not yet been verbalized is another. We’ve all done a little of this before, and the demand for it is only going to increase. - social implementation
This page is auto-translated from /nishio/まだ書かれていないことの森を切り拓いていく仕事 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.