About AI Learning Tools

First of all, I believe that as long as the note-taking method and mindset are correct, any note-taking software can be used. The most important thing for making good atomic notes is that the software must have the bidirectional linking function, because we need to build connections between knowledge.

Now let me talk about tools like IMA and NotebookLM. These are AI knowledge bases. How do they work? You throw in the books and videos you’ve found, and let the AI study them for you. It goes through all the material, summarizes and extracts the key points, then produces notes or even generates mind maps for you. I don’t recommend relying entirely on this method to manage your knowledge, because the learning process isn’t done by you, and the notes produced aren’t created by you either.

Here’s a concept: cognitive debt. When we use AI tools to think in place of our own brains, our efficiency improves in the short term, but in the long run our cognitive abilities weaken significantly. Our capacity for thinking and problem-solving will deteriorate.

So, we still need to manage our own knowledge—organize and learn it ourselves. That’s where Obsidian comes in: it’s our own personal knowledge management software.

An ideal workflow would be: use NotebookLM or similar tools to let AI summarize the knowledge points from books and videos, reducing some of the burden of reading them yourself. Then, treat the AI’s output as study material, and make your own notes from it.

Another concept is desirable difficulty: when learning, we should keep the difficulty level within an optimal range. If it’s too hard, we won’t understand and may give up. If it’s too easy, it won’t provide any real training.

When making your own notes, don’t just copy and paste. Use your own words to express the ideas. This is an application of the Feynman learning technique—input, then output.