Author: danny

A friend asked me why it seems like I know about everything or in every field? Aside from some previous experiences or current projects, actually many times, I learn and apply things on the spot. Today, I will share with everyone how I use AI tools and Notebooklm to embark on the self-learning journey of an ordinary person.

First of all, I want to say that this article is aimed at systematic and structured learning and understanding of a specific field/thing/concept, and building your own knowledge system and map. If you only need to have a slight understanding of some concepts and know what this xx is, then just asking the mainstream AI on the market may be sufficient.

Using AI to learn and understand a new thing currently has several bottlenecks and limitations:

The first is hallucination; AI (likely) will give you some fabricated data and information, especially in niche fields, due to insufficient corpus and learning materials;

The second is the lack of numerous details; due to copyright issues, AI won't read through entire articles or books by itself, and training materials are generally reviews and comments from others, especially information in niche fields is particularly scarce;

The third is the inability to accurately describe problems; assuming you have not previously come into contact with this topic, you probably can't describe the questions you want to understand well, don't know the causes and consequences of these matters, and even less so can systematically and structurally collect information and form a systematic learning framework.

Theoretical Part

My approach is quite simple: utilize the academic "citation (quote/reference/impact factor) network" to purify information, and then use AI to argue and brainstorm in a "mutual combat" of left and right brains to structure an understanding of a new thing.

Streamlined Workflow:

Find valuable papers - Upload to Notebooklm - Use AI tools to generate prompts - Ask and learn in Notebooklm - Supplement valuable papers into Notebooklm - Learn in Notebooklm - Repeat this process

Complex Workflow:

Step 1: Follow the clues (Time: 0.25 hours)

Don't search for "What is XX, what is the principle of this?", but directly look for the "pillar" of that field.

1. Call AI (Gemini / Perplexity): Directly ask: "In [specific field], who are the three recognized top authorities? What are the 1-3 highly cited classic papers that laid the foundation for this field?" (For example, in the LLM field, focus on Attention Is All You Need, etc.). It represents the "present life."

2. Download first-order literature: Extract the references of these 1-3 core articles and download all core literature they cited. It represents the "past life."

3. Refine high-frequency second-order literature: Cross-reference the references in the first-order literature to filter out the top 10 most cited articles and the top 5 most frequently occurring articles. This represents the "later stage."

Core logic: Looking at the world through the eyes of masters is the least costly shortcut. Don't underestimate this step; what you download is the core thought evolution map of the field over decades.

Step 2: Build a structured knowledge base (Time: 0.25 hours)

Upload all classic literature filtered out in the first step at once to Google NotebookLM.

Generally, as long as they are classic articles, these two sources are enough:https://scholar.google.com/ or https://arxiv.org/

Why NotebookLM? Because it never produces hallucinations. It only answers questions based on the materials you provide.

Through rigorous literature screening, you have cut off junk information from the internet, establishing a pure and highly focused knowledge base for this field.

Step 3: Mutual combat between different AIs (Time: 1-3.5 hours)

This is the core of the entire workflow. You let different AI with varying characteristics cross-examine in your knowledge base, forming structured knowledge paths and logical deductions, ultimately developing your own insights.

Ask active questions instead of passively learning. Active questioning (interest) promotes brain thinking.

1. Find anchor points: Ask Claude, Deepseek, Gemini, or Perplexity: "What are the core controversial issues and underlying theoretical frameworks in the field of xx?"

2. Close-loop questioning: Take these core controversies back to NotebookLM and ask: "Based on the literature I uploaded, how have the masters addressed these core controversies? Please provide specific literature sources and reasoning logic."

3. Dimensional reduction analysis: Copy the rigorous answers generated by NotebookLM and throw them back to Gemini or Claude, which have strong logical analysis capabilities. Give instructions: "Please critically examine these viewpoints and point out any logical flaws, limitations of the era, or blind spots. Based on this, what 3 deeper questions should I continue to pursue?"

4. Cognitive spiral upwards: Take the flaws and new questions pointed out by AI and return to NotebookLM for answers.

Practical Application

Let me use "What exactly are LLMs (large language models)?" as an example ?

Step 1: Follow the clues (Time: 0.25 hours)

I simultaneously asked Gemini and Claude - "hey, you really did it, the answers they gave

gemini

claude

Then you suddenly remember that your middle school teacher said that scientific theories must connect the past and the present, having a past life, present life, and later stage. So you let AI help you research which papers these core articles have referenced (usually found in the "literature review"), as well as which later articles cited the core articles, letting AI filter it out for you.

Step 2: Build a structured knowledge base

Due to some original LLM characteristics and AI permissions, we need to manually download (or you can have your lobster do it)

Generally, https://scholar.google.com/ and https://arxiv.org/ are completely sufficient.

You will download and then place it in Notebooklm (which currently supports around 300 articles in one library).

Step 3: Mutual combat between different AIs

You can first ask some simpler, intuitive questions in Notebooklm, then discuss and explore your understanding with other AIs, and afterwards send the conclusions back to Notebooklm for it to refute, argue, supplement, and correct.

Notebooklm's answers and comments:

Repeat this process several times until you can organize your own mind map.

Then, if you want to be hardcore, you can ask Notebooklm to give you a quiz to test yourself.

By now, you have a certain understanding of this field (at least knowing the past life, present life, and later stage, when others ask you, you can speak for an additional 5 minutes~)

Postscript

Keep your "knowledge base" saved (and updated in real-time, you can have your lobster do it), create a separate folder - for example, I have a collection of theoretical articles related to "contract trading." When analysis is needed, simply open this folder, describe the data and cases, and you can basically conduct an "analysis without hallucinations."

It's not that current AI models cannot perform deep thinking and analysis, but rather that you haven't used the right tools. (One crucial parameter in LLMs is the constraints and input conditions)

Using AI is one skill, but how to empower humans with AI is another skill. Using AI is one ability, but how to make AI empower humans is another ability.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。