December 1, 2023
We read them for pleasure and to learn. Authors write them to transfer knowledge across decades.
The rise of artificial intelligence points to a massive shift in the publishing world.
It’s not just that generative AI makes books easier to write. People are using AI to read and summarize them as well. That trend was noticed by IEEE Senior member Roberto Saracco, the writer behind the IEEE Future Directions blog, who noted in a pair of posts that AI-assisted writing tools and self publishing are turning millions of people into authors, and that the rise of AI may change the way we interact with books altogether.
We talked with him to illuminate where these trends are going.
How do you think AI summarization and question-answering tools are affecting traditional book reading habits?
We are in the very early stages of AI usage for book reading. In three to five years, the use of AI as a reader assistant is likely to become much more widespread. ChatGTP-4 Turbo will become available to the mass market sometime in 2024. It has the capability to digest the equivalent of a 300-page book, and even larger one in case of narrative. I would expect e-readers to support this feature seamlessly, as they are supporting voice reading today.
I see two converging forces that will result in a more generalized adoption: on one side students will find it very useful to have summaries and the capability to have questions answered. On the other side I see companies leveraging these capabilities both internally and as a way to provide conversational support to their users.
People read fiction and popular non-fiction for pleasure. What’s the point of an AI summarization tool for those readers?
There is a significant number of people using audiobooks, in spite of the pleasure given by direct reading of text. Audiobooks have expanded the fruition of books, they did not substitute the reading. Likewise, the ability to summarize will make more books come to life. The possibility to interact conversationally with a book will open up a whole new way of fruition.
With AI-assisted writing tools enabling more people to publish books, what do you believe is the impact on the quality of published content?
We already have the answer today. There are millions of published books. Most of them don’t come through the classic channels; think about ebooks and blogs. Most are of low quality, yet a sizable subset fulfill the need of the long tail. This trend is just going to continue. It is already happening for music. There are millions of music authors, thanks to tools that make music creation and publishing easy.
What are the educational implications of students relying more on summarized content provided by AI rather than reading full texts?
Many textbooks are boring. You may remember spending your time underlining, highlighting paragraphs and words. Isn’t that a form of summary? Now it can be done in more effective ways. Besides, students will need to learn how to exploit summarization and questioning of books.
Education is not about throwing a book to a student. It is about coaching the student to extract value from books. I feel learning will be more effective since you move from a laid back approach, which we can call passive reading to a lean-forward approach or active reading. And today there are so many books that you cannot read them all. But possibly, through AI, you can get a glimpse of many and through questioning you can learn.
Do you see any benefits to the idea that AI can make it easier for more people to become writers?
The pencil has democratized writing. This did not produce better writers but it provided the potential to many more writers. From this larger pool of writers we got, statistically, more pieces of high quality writing. So yes, I see a benefit. The offerings will keep growing and amidst the many pamphlets, there will be some worth your attention. There is also another, even more important benefit. By enabling more people to write you are stimulating their intelligence and their thinking. In the process, we are creating a more knowledgeable society.
In one of your blog posts, you note that by the end of this decade, machines may be the primary readers of IEEE articles. And we can assume that will be the case for a lot of written material. What do you think that means for the way people live and work?
I am talking about personal (or company) digital twins that will be able to transform knowledge into executable knowledge. By that I mean knowledge that can be used here and now, either by becoming aware of it or using a machine to carry out a given task.
This isn’t offloading the responsibility to know things to a machine. Each individual will remain the ultimate responsibility for the application of that knowledge, hence I will need my own knowledge on how to manage that machine knowledge. When a construction engineer is designing a building using a tool to calculate the need for pillars and beams, all the calculation is done by a tool, but the engineer has the ultimate responsibility.