tokoroten As I touch ChatGPT’s GPT4, I’m keenly aware that the world of this is seriously coming. I guess from now on, finding issues and verbalizing issues will be the job of humans. Yoichi Ochiai [Revitalization Strategy for Japan image

nishio I was thinking about what book will look like in 5-10 years, and from this point of view, books are mass production. I wonder what will happen when it changes to Diversity Production.

  • relevance - Observations on the Publishing Industry - People who think for themselves want a list of facts and information.

  • tokoroten I’m sure books will be personalized and summarized by an AI that estimates the knowledge network model in my brain!

  • I think amazon is going to create an AI or something that will search for appropriate knowledge from the library.

  • This will increase the number of people buying books.

  • tokoroten I guess the next step will be the age of buying books to make personalized AI smarter…

  • hrjn It won’t change in terms of producing books, but it will change the way people read them. After all, we can’t handle information that doesn’t exist in this world in any way, so we have to write it.

  • Or maybe you could just write the facts as you see fit and it would autoplay the intro (or a summary of related research) and a summary or so.

  • hrjn In other words, there is one chance that the act of writing a book will be reduced to ā€œnovelty (patentability)ā€ only and incidental information will be fleshed out as appropriate. It may be.

  • yhara Ah, so it could be something like ā€œa book with the same content, but with varying degrees of annotation depending on the reader’s prior knowledgeā€?

nishio Many people may think of writing a book as writing, but I checked today and my Scrapbox is 58 times larger than the ā€œtechnology behind codingā€ It’s a lot of volume, and in the writing process, I output more and then cut it down to complete it. nishio Essentially, it’s a question of ā€œwhat order to put the knowledge that’s on the grid in a single strokeā€, ā€œB should be first to explain Aā€, ā€œBut because of its relationship to other things… Then B should be firstā€¦ā€, ā€œIt would be more efficient to repeat a detailed loop, but the whole thing would have to be divided into a dozen or more chaptersā€¦ā€, ā€œThe depth limit of the headingā€¦ā€œā€¦ - Knowledge Network - Books are a tool for finitization - Books are already a product of abstraction The hard part is to ā€œput yourself in the shoes of readers who don’t know the detailsā€ and ā€œformat it in a way that makes it easy for them to read,ā€ something you already know ā€œin detail enough to write a book,ā€ and LLM can play the reader’s role. - It’s hard for one person to play two roles.

nishio I’m a little fuzzy on whether classic and preprint are two ends of a barbell if we limit it to the activity of ā€œreadingā€. Old enough sentences are swallowed as LLM training data via the Gutenberg project, etc. The edge of copyright, ā€œa system whereby old humans get in the way of AI to protect their vested interests,ā€ may be the edge of the barbell.


This page is auto-translated from /nishio/ę›øē±ćÆćƒžć‚¹ē”Ÿē”£ 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.