Human memory is fragile. Therefore, personal notes have become an important means for us to build our second brain and carry out knowledge management. However, the rise of artificial intelligence has undoubtedly changed the way we process information and knowledge. As an "efficiency addict" and "software enthusiast" who has been paying attention to note-taking methodology and AI software for a long time, I deeply feel the impact of this change. After the arrival of AI tools headed by ChatGPT, I once wanted to give up taking notes, but I couldn't find a reason to continue to take notes. Subconsciously, I think that the positioning of note-taking has changed in this era, and my thinking and habits have not adapted to such changes, and I am very confused:
- In the context of artificial intelligence (AI), are our previous note-taking strategies still valid?
- Where should our knowledge graphs, bidirectional links, and those carefully crafted "second brains" go?
Old Note Mode: Miner Mining
Previously, we learned much like miners digging for precious ore . We are like them going deep into the mine, digging patiently, finding precious ores, and then refining them into valuable wealth. Through reading and learning, we excavate and screen from the massive knowledge, and finally form our own knowledge system. Our mind is the rich mine, the notes are our miners' tools, the double chain is the map we use to find the mine of knowledge, and new knowledge is like a piece of gold found by ancient miners in the unknown cave. .
Site Selection: Finding Knowledge
Just as miners select sites for construction sites and look for mines of great value, we look for areas of interest or usefulness in the ocean of knowledge. What this process involves is that in a large amount of knowledge, we need to find which ones we really need. At this stage, we need to read a lot of books, articles, web pages, listen to lectures, and practice in person, in order to find valuable "mines" from the endless content of the Internet .
Prospecting: Quick Reading and Outlining
This is like looking for a gold vein. Once the "mine" is found, we have to conduct exploration to find the gold vein and determine the best direction and path for digging . This is equivalent to our process of searching and filtering relevant information. When we learn a new topic, we usually start with the most basic and superficial content, use search engines or consult encyclopedias and other tools for the first step of digging, first determine the basic concepts and framework of this field, and find out what belongs to this field. The "mineral lode" of the mountain. But we often don't know where to start, and we're not sure where to learn. The learning path is extremely unsystematic and it is difficult to grasp the whole picture.
Excavation: Excerpts and Records
Through a macroscopic understanding of the entire field, we have found the most suitable entry point. Then, we use note-taking software (such as Notion, Evernote, Roam Research, etc.) as our tools to record our thoughts, extract and record important information, and collect valuable "ores" just like mining gold veins. We store these "ores" in our own "baskets" for later retrieval and use. But such mining is inefficient and highly repetitive. We will repeatedly organize similar content, but the gains are limited.
Screening: Organizing and Connecting
Just like miners screen out valuable ores, we also need to classify and process knowledge, remove impurities, and keep valuable content. At this stage, we need to organize knowledge and establish a knowledge system through functions such as tags and double chains of the note software, as well as note methodologies such as "card notes" and "PARA", so that the knowledge we have learned becomes more structured and easier. understand and apply. This process requires our brains to analyze and judge the importance and relevance of information, just like miners discerning the value of ore. We will obtain a large number of fragmented knowledge points in this process, but it is difficult to effectively integrate them.
Refinement: Forming new understandings and insights
The final step is refining, just as miners refine raw ore into pure metal. We need to internalize knowledge and make it our own. This process requires us to think, practice, or communicate with others, so as to form our own opinions, establish a double chain of associations between knowledge and knowledge, and form the crystallization of thinking. However, the external knowledge network we build is not as comprehensive as the neural connections in the brain. Information retrieval is inconvenient and easily forgotten.
Note mode in the age of AI: machine mining
After the arrival of AI, looking back at the entire learning workflow that I am "proud of", I realized how primitive and inefficient it seemed. Like mining, the process is inefficient and labor-intensive . Moreover, the process of building and refining knowledge is limited by individual time and energy. What's more, dealing with the ever-growing amount of information and knowledge becomes a major challenge. Every day we are confronted with a massive amount of new information, which is even more difficult to process with traditional methods.
However, the emergence of AI tools has made knowledge acquisition and management more convenient and efficient. Just like the introduction of advanced mining machines in the industrial revolution , AI can quickly go deep into mines, excavate on a large scale, and improve the efficiency of ore acquisition.
- In the "site selection" stage, AI is like a " local person " who is very familiar with geography , directly pointing us to the location of the mine. It can help us quickly find areas of interest through the semantics of our search more intelligently, and even give suggestions for in-depth exploration. For example, New Bing, which integrates GPT capabilities, has shaken Google's dominance in the search field to a certain extent.
- In the "exploration" stage, since AI absorbs most of the content on the Internet, it can organize the knowledge of a field from a very macro perspective, helping us quickly get an overview of this field and the most important reference materials. Just like directly providing a " treasure map " of a mine, we can directly access valuable knowledge without having to explore by ourselves.
- In the "excavation" stage, AI is like a large " automatic excavator " that can help us dig quickly and in large quantities. For example, models such as GPT-4 can quickly extract useful knowledge from a large amount of information, which greatly improves the efficiency of our knowledge acquisition. These tools can automatically read, understand, and summarize text, saving us a lot of time.
- In the "screening" stage, AI can automatically filter out the most valuable information to us based on our preferences, just like an excavator automatically separates ores and impurities . AI uses technologies such as embedding to enable us to automatically perform content association and reasoning without manually establishing complex double chains. We can quickly find the most semantically appropriate content in a huge information base, help us classify knowledge, save time in sorting out, and transform knowledge in the most understandable and memorable form.
- In the "refining" stage, AI can help us refine the content like a furnace. Through dialogue, it can help us understand knowledge from various angles and answer our doubts, thereby assisting us to form our own insights and insights.
Now, we can organize and manage knowledge more conveniently, without having to manually build complex knowledge networks. At the same time, AI can also better structure the content and reduce our excessive indexing and duplication of information. And all of these are precisely the things that I have spent a lot of time and energy doing before.
In the age of AI, what do notes mean to us?
I have spent a lot of time and energy honing my note-taking ability and skills, just like miners carefully polish their tools and learn mining skills. However, when I saw that AI can easily and high-quality complete these tasks, I feel that all previous efforts have been wasted, just like I drew a mineral map (star chain map of notes), only to find that AI directly gave me a more detailed, Treasure maps that make more sense and have a wider range. Those note-taking tools and skills that were once proud of seem to be less important. Under such a "dimensionality reduction blow", I doubted my previous note-taking method for the first time-was all my previous efforts in vain? Will notes be replaced by AI?
When the tide receded, I realized that "knowledge management" was not so necessary
The advent of the AI era has gradually extinguished my enthusiasm for notes. Comparing the time I put into various cards, double chains, and how it really helped me, it really doesn't seem so necessary. Most of the time, I was just indulging in a game called " Building a Second Brain " , taking notes for the sake of taking notes, and being enslaved by tools.
At this time, I suddenly remembered a meme:
The debate over the necessity of note-taking software has existed long before the arrival of AI tools. Most of the people in the middle are struggling with how to build their own note-taking system, except for those with the highest and lowest IQs. People with high IQs are good at internalizing knowledge, which is equivalent to having a knowledge management system in their minds, and only need to record some simple notes. Ordinary people do not have such a powerful brain, so they can only rely on some note-taking software to help them build a "second brain" to complete knowledge management.
At this time, I suddenly understood that the reason why I feel so frustrated is because my personal knowledge management ability is so worthless in front of AI. If unbalanced knowledge management capabilities are the cause of this distribution, then the intervention of AI can basically smooth out this imbalance. The importance of note-taking software will be weakened to a certain extent. Thanks to the assistance of AI, most ordinary people can also have an intelligent knowledge management system to help them organize and find knowledge. People's differences in knowledge management will become smaller and smaller, and finally everyone only needs an extremely simple note-taking software, which can record and find knowledge very easily.
The core of notes lies in the flash of thinking
Does this mean that all kinds of software based on bidirectional links and knowledge graphs will disappear? I don't think so. Because just like machine mining cannot completely replace manpower, AI cannot completely replace our learning and thinking. The double chain in note-taking software can not only help us establish the association of objective similar knowledge, but also include some superficially irrelevant but subjectively related knowledge (understanding, insight and perception). These flashes of thinking are the core of notes, and they cannot be replicated and achieved by AI.
We can find that although AI has been able to summarize and classify, its ability to generate content is still relatively poor. The content written is very "plain water", broad and dull. As one of my favorite metaphors puts it:
ChatGPT is a blurred JPEG of internet content. —— Ted Chiang
AI-generated content is essentially a highly compressed image of existing content on the Internet. This means that its content is destined to be very blurred, many details are lost, and it is not sharp . And those subjective insights are often new ideas that are sharp enough to pierce our solidified thinking, which cannot be replaced by AI .
For example, when I saw a news article about AI products exploding, it reminded me of another article I read in the past analyzing the reasons for user growth. This inspired me to have new ideas on the guiding means of product interaction. I will document and relate to this key insight.
Therefore, we can use AI to obtain some existing information to stimulate our thinking. But you can't let AI completely replace yourself for analysis and thinking. This will only make your thinking more and more lazy and solidified, and you will not be able to create original content.
Therefore, I think that AI can replace some passive and objective knowledge sorting functions in note-taking software, but active thinking and insight-related functions will not be replaced, and we still need to complete it ourselves. AI can free us from mechanical knowledge summarization and organization, and return to the essence of the note-taking system-record thinking and generate inspiration.
Embrace the AI era and transform note-taking workflow
As with past industrial revolutions, we need to accept this new reality and figure out how to adapt to it. In the age of AI, we need to transform the previous note-taking workflow and combine new technologies with past achievements.
Now, I split the function of notes into two parts: objective notes and subjective notes .
- Objective notes are undertaken by AI: such as names, citation links, article summaries, etc. The mechanical work of such objective links and knowledge sorting is handed over to AI, which can save us time and energy.
- Subjective notes still need their own: such as the inspiration triggered by a note, the association and insight between two notes, etc. These subjective insights require us to think, and it is difficult for AI to replicate this process. Therefore, we can devote more time and energy to comprehension and inspiration recording.
Be the "Mine Master" of your own knowledge system
According to this strategy, I used the capabilities of AI to upgrade my "mining" note-taking workflow. Mainly replace the previous relatively objective work with various AI tools. Now, I can just sit back and use my brain like a " Mine Master " .
Site Selection: Finding Directions
In the stage of site selection, the biggest difficulty is that we often don't know where to start, and we are not sure about the direction of learning. Therefore, I mainly use some search engines integrated with AI capabilities to help me quickly locate specific areas and the direction I want to explore.
New Bing Search
In the past, we could only find superficially relevant content through a method similar to " keyword matching ", and then find deeply relevant content from it. However, often we don't know which keyword to use to search (uncommonly used synonyms, or expressions used by some laymen), which causes us to miss the content we want.
Now, with the GPT-based New Bing, I don't have to care so much about how "perfect" and "professional" my keywords are. Because it can find the most relevant content by analyzing my semantics , and can also correct its own prompt through multiple rounds of dialogue, which greatly saves my retrieval costs.
Metaphor search
In addition, Metaphor is also an AI-based search engine. Given a prompt, it finds links that are most likely similar to that prompt. Through it, I can find the learning resources and high-quality content I need most through natural language descriptions. Compared with New Bing, Metaphor is more suitable for me to use when I have a more detailed search tendency, and sometimes I can find some relatively unpopular websites that are very relevant to the content I want, which brings me some surprises.
Exploration: Quick Start and Overview
As the saying goes: "The master leads the door, and the practice is up to the individual." But most of the time, we don't have the role of a master to lead us in. I always look for some tutorials or shares before, and follow the path of these people. But a person's vision is still limited, and the path they take may not be very reasonable. It is often only halfway through that they find that their purpose is inconsistent with their own, or the path they are taking is also wrong.
Well, since AI has a breadth of knowledge that we cannot match, it is the best guide . Therefore, at this stage, I mainly use AI's planning and information gathering capabilities to help me tailor a learning plan.
aomni
Aomni, an application similar to AI Agent, is like a mentor or a senior, making a very detailed study plan for me and helping me complete research in a field. Through the key papers and articles it helped me find, the content I read can be very streamlined. In addition, it will also give me some additional "admonitions", such as the use of tools, strategies for community and follow-up continuous attention, etc.
Excavation: Excerpts and Records
Extracting and recording summaries from existing text passages is a very difficult and slow task for us humans, but it is a piece of cake for AI. This part is also the fastest-growing and most mature function in the field of AI. Various note-taking and reading software have integrated such functions into their own products, allowing us to easily inspect and read and quickly understand the core content of the article.
Readwise Reader AI Assistant
Automatically generate summary in sidebar
Among them, Readwise Reader's Ghostreader is one of the AI functions I use the most. Because of my procrastination, previously saved articles have been hoarding in my reader for a long time. With this function, I no longer need to read word by word, but can rely on AI to assist me in reading, quickly understand the main points of the article, and then decide whether to read in depth.
Claude 2
Also, I recently discovered Claude 2. Thanks to its 100k long contexts, I can just drop a paper or even some short books in and get a fairly high quality summary in seconds. Unexpectedly, the "quantum speed reading" that was ridiculed by us a few years ago is now truly realized on Claude.
Screening: Organizing and Connecting
The screening and organization of knowledge is similar to the screening of ores, and has a certain way of judging and organizing framework . And my previous note-taking methodology finally came in handy at this stage.
Prompt template for note rules
Using ChatGPT's powerful prompt understanding ability, we can organize our own note-writing methodology into text rules, tell GPT to make it like a machine, and convert all the content we feed it later in the same way. .
For example, based on such rules, I feed it a piece of text that I recently bookmarked while reading, and it automatically converts the content into Markdown format according to the rules. I can simply copy and paste the content to my own Note software.
Notes in Markdown format generated by GPT
Paste directly into note-taking software
Refinement: Forming new understandings and insights
If we talk about the previous relatively mechanical and passive note-taking operations, AI can basically do it. Then when we reach the stage of refining knowledge and generating insight, it is time for us to think actively. As mentioned earlier, AI-generated content is abstract and vague. All the previous work was to improve the efficiency of notes, so we can accept the fuzziness in the content generated by AI. But if we want to form a real crystallization of thinking, we need to integrate these contents into our cognition and memory, collide with our original views and output them.
In this process, the most ideal role of AI is a listener. Listen to our point of view, and give us some relevant guidance and inspiration, let us refine our own understanding in the output of the point of view.
Pi.ai
For example, Pi.ai is currently the product that best matches my vision of an AI-inspired assistant. Different from the objectivity and coldness of ChatGPT, its conversation is very elegant, and its personality is very pleasing. The interaction of the entire interface is also very soothing and natural. There will really be a sense of déjà vu to have a long conversation with a good friend. Moreover, it will throw a question after each paragraph, which attracts me to communicate with it constantly.
Whenever I have a certain opinion on a point of view, I will enter the mode of " help me think about something clearly " and have a dialogue with it. According to the questions it raises, keep throwing out your own ideas, gradually dig into the essence of the problem, and finally realize the epiphany on this topic.
Through the above workflow, I will hand over the inefficient work of summarizing, classifying, sorting, and searching that I am not good at to AI. Spend more of your time and energy on deep learning and understanding new knowledge, generating new ideas and insights. No longer constrained by the workflow and methodology of notes, no longer enslaved by notes. Instead, I use the ability of notes and AI to allow me to better capture those precious moments of inspiration. Finally, I completed the transformation from "miner" to "mine owner".
Future: Symbiosis with AI
In the past few months, " AI replacement theory " has been heard endlessly. But it seems that for most people, there is little change in work and life. I think the same is true for note-taking software. AI won't completely replace anything. It will just gradually fit into our existing workflow, displacing the inefficient, mechanical parts of it. And those that cannot be replaced by AI are the essence of these things, the real gold and silver extracted from the ore.
So, I don't think it's necessary to be hostile to AI and worry about how it will replace what. Might as well try to embrace AI, live with AI, make use of its characteristics, understand its limitations, and at the same time maintain our own creativity. In this way, the convenience brought by AI can be enjoyed as much as possible, and the old workflow can be given new life.
If you have other related feelings and experiences, welcome friends to discuss!