Mastering Difficult Books with Google Gemini Part 10: AI Reading Methods to Engage Your Brain

Hello! I am Toshio Tsukuda, a poet who delivers tips on how to use Gemini in documents with the motto "Boost your reading productivity with AI."

Dependence on AI causes the brain to decline.

In the previous article, I shared that shocking news from MIT. In this article, I will discuss reading methods that protect and activate your brain, based on those experiments and insights from neuroscience.

And then the mind became empty

Are you familiar with the term "Cognitive Load"? It classifies the mental effort required to proceed with learning or problem-solving; simply put, it is the load necessary to advance thinking. There are three types of cognitive load:

1. Intrinsic Cognitive Load (ICL) This is the load caused by the complexity of the learning content itself and the learner's prior knowledge. It is the essential mental effort associated with learning or problem-solving.
2. Extraneous Cognitive Load (ECL) This is the cognitive load created when trying to receive and understand information that has noise in its design, organization, or editing. It refers to non-essential mental effort. It is a "bad load" that hinders learning and the like.
3. Germane Cognitive Load (GCL) This is the beneficial mental effort spent on building and automating schemas (knowledge frameworks) that support learning. It is a "good load" that nurtures and trains our brains.

Excessive cognitive load, especially extraneous cognitive load, can hinder schema acquisition and reduce the efficiency of learning and problem-solving processes. The MIT paper points this out.

Also, if a state of high extraneous cognitive load continues for a long time, it becomes stress and damages the hippocampus (the part of the brain that controls memory). As a result, the efficiency with which the hippocampus processes information decreases. Speaking of stress, operational loads such as complicated tasks are also a source of stress.

In that sense, using an LLM (ChatGPT in this case) reduces extraneous cognitive load. However, according to the aforementioned research, the paper points out that the reduction in germane cognitive load, which is beneficial for learning, is significant, potentially leading to lost opportunities for schema construction and a decline in critical thinking skills. It is quite a difficult matter.

Making AI an external hippocampus

Information carried into the brain through the five senses, such as sight, hearing, and taste, is temporarily stored in the hippocampus's working memory. The information is then organized, transferred to various departments in the brain, and inscribed as long-term memory.

The amount of information that can be carried into the hippocampus is not infinite, and it is lost once it exceeds a certain amount. Many people have experienced their computer's memory reaching 100% and stopping. It is thought that what we call "not being able to think" is the same state as this.

The two types of stress I mentioned earlier also add to this, putting pressure on working memory. How to secure space in working memory—that is important in the sense of protecting the brain and making it serious.

Therefore, we take preemptive measures so that the brain does not end up in such a state. We make AI like an external hard drive (hippocampus) for the brain, artificially increasing working memory.

By doing so, we create space in the brain's memory and intentionally create a state where new ideas and insights are more likely to be born (= emergence). This is the goal of this reading method. Reading with AI can be said to be like correcting poor eyesight with glasses.

Create a dedicated commentator for the book you want to read using AI

To maximize the results of reading using AI, optimizing the "AI as a speaker" will increase the effectiveness of your reading. For example, if you are going to read a philosophy book, enter the following prompt into the AI:

"You are the world's greatest philosopher. I am going to read a book called [Book Name]. If there is anything I don't understand or have questions about, I will ask you, so please answer."

This is a technique called "persona setting" in prompt engineering. When you set a "persona" suitable for explaining the book you want to read, the AI will behave accordingly, gather information, and prepare more optimal responses to your questions.

What you should be careful about is not to set the book's author or a real person as the persona, for example. The AI will try to live up to expectations and create plausible answers rather than correct ones. There is a possibility that what is known as hallucination will occur.

Rephrase the AI's answers into your own words

When the AI responds to your question, try to reply using words that come from as deep within yourself as possible, saying something like, "Does this mean such and such?" You don't need to try to sound clever. Just reply as if you were writing down what you felt at that moment in your reading notebook.

Of course, summarizing in your own notebook or writing in the book is also effective. However, when you use AI, it acts as a teacher grading your work.

When studying on your own, your understanding often becomes idiosyncratic, and you frequently end up remembering things incorrectly. Since LLMs are generally eloquent, they can make you feel like your need for approval is being met, which makes it easier to increase your motivation for learning.

If you are not convinced, please feel free to debate it.

If you continue this practice, your logical thinking should improve. This also leads to critical thinking skills, which in turn can lead to insights that even surprise you.

"Does that mean this doesn't happen?"
"I thought about it this way."
"That idea is similar to this one."
"I still think that's wrong."

Whatever it is, try to keep a rally of questions and doubts going, starting from the book. As you get used to asking questions, your problem-solving skills will also improve. You might think that debating with AI sounds difficult, but it's okay.

The good thing about AI is that it talks to you according to the "amount of knowledge you possess." If you are a junior high school student, it will talk to you using analogies that a junior high school student can understand.

Therefore, you don't need an obscene amount of knowledge or mental strength. The reason I can challenge slightly difficult books is not because I have a lot of knowledge, but because I use AI as an external brain to expand my own thinking ability.

Thoroughly reducing "bad load"

It goes without saying, but you cannot pour more water into a glass that is already overflowing. If you want to use the glass, it is important to create space for the water.

To keep your brain working energetically, it is important to create space in your working memory, which is filled to the brim with information and stress. In other words, controlling the load we are under leads to making the brain work seriously.

The first thing to do is to reduce extraneous cognitive load. This is where the triage memo created by Gemini and NotebookLM comes in handy (for details, I will leave it to this article).

Setting off on an adventure without a map is quite a heavy load, isn't it? That is why, before reading a book, you should create a map of where the treasures are and where the monsters appear. It is, so to speak, the "visualization" of reading.

Something you should do in conjunction with this is "sorting." By setting a mindset such as "What is my purpose for reading this book?" and "What information do I want to get from this book?", you can sort the vast amount of information into "necessary/unnecessary."

Dividing "essential load" into manageable sizes

Once you have finished weeding out the extraneous cognitive load, try working on sorting the difficulty that the book originally has, in other words, the "intrinsic cognitive load." The criteria for sorting here is "understand/don't understand."

Reading is sometimes called "having a dialogue with a book," but try adding an AI to that dialogue. Set a persona and keep asking questions to the AI optimized for that book, saying, "I don't understand this part."

Then, like peeling an onion, the "amount you don't understand" will decrease. If you do that, the intrinsic cognitive load will change into germane cognitive load, which is a "good load." Once you break it down into a "just right, slightly difficult" level where you think, "I might be able to do it if I try a little," all that's left is to enjoy it.

The important thing is not to overdo it. Originally, this reading method, "Reading Journey," is an emergent reading method aimed at gaining knowledge and insights from books. So, in the extreme, it's fine if you read 5 pages and finish, or read for 10 minutes and finish, or even if you don't finish the book. Receive stimulation from the book and from the dialogue with the AI, and build a mountain of treasure in your heart. Let's enjoy it with that image in mind.

The book I am reading as core reading this time.
(I am participating in the Amazon Associates Program.)

Thank you for reading until the end.

I am always encouraged by your likes, follows, comments, and tips.
Thank you so much!
Really, thank you!

[I also write articles like this]

#33DayWritingChallenge