Author: Yaki
I recently came across a Douyin creator discussing learning and reading. He seriously explained “don’t read unless necessary” and “interactive learning,” not in a way that creates anxiety, but by using logic to explain why they make sense.
The gist is: Benjamin Bloom’s 2 Sigma effect/problem (one-on-one tutoring and mastery learning) can make learning efficiency surpass that of 98% of students!
Bloom’s exact words: "the average tutored student was above 98% of the students in the control class."
But the drawbacks are also obvious: this coach-style, one-on-one questioning approach to learning places extremely high demands on the coach, and is a huge challenge in terms of time, energy, ability, and money.
Now that AI is available, AI’s mastery of knowledge far exceeds that of humans, while the cost of using AI is extremely low. Could AI be used to carry out “2 Sigma” coach-style learning and improve one’s learning efficiency?
Driven by this curiosity, I began learning how to set up Claude Code by following the creator’s videos and started experimenting. In total, I tried two different methods~
Method One: Have AI Help Me Understand a Book
Without using any prompts, I asked Claude Code to help me understand a book. I chose 《Guiguzi》, and after waiting about 5 minutes, it mapped out the structure and framework of the entire book.
It is worth noting that the reading order organized by AI differed from the order of the table of contents. It told me that I should first understand the structure of 《Guiguzi》, “Baihe” is the underlying way of thinking that runs throughout the bookYou need to read and understand this section repeatedly first. Only after you’ve fully grasped it should you move on to reading the next section of the structure.
This also reminded me of another knowledge blogger, “Xingzhe,” who said something similar: you don’t necessarily have to read a book from beginning to end in the author’s order. Instead, start with its structure, identify the book’s categories, and read through one entire category at a time. On this point, AI and I are in perfect agreement～
Method Two: Coach-Style Learning (Prompt Version)
I also started trying coach-style learning using the prompt from the video. Its learning process works like this:
- First, it generates the first document.
- After you finish reading the document, it asks you a few questions.
- You need to write down your takeaways from the document and any questions you’d like answered.
- Based on what you understand, it generates a second document, and so on.
- Topics you already know or understand will come up less often. Your questions and any takeaways that aren’t clearly explained will be revisited over multiple rounds until you fully understand them. Only then will you move on to studying the next section.
It’s very easy to get into a state of “flow” this way. It really feels like a knowledgeable coach patiently explaining what they know to you. When a confusion is cleared up, that feeling of satisfaction is like the endorphin rush after solving a long-standing problem—and now you can get it quickly just by having a conversation. I can’t help but marvel: AI really is powerful.
On “Only Read When Necessary”
That said, I don’t have that “only read when necessary” mindset. I’d describe it as a particular attitude, perhaps driven by a desire for traffic or some kind of provocative expression. Of course, that’s not important!
For me, discovering new information to satisfy my curiosity is a pleasure, so I never tire of exploring. But for other people, exploring information may not be necessary.
Do we really need to know so much?
Is it really necessary to know so much?
I believe everyone has a different answer.
I hope today’s post was helpful. I’m Yaki Nie～
Good night 🌙