The Scam Called “You Don't Have to Remember Anything” • Zettelkasten Method
The Scam Called “You Don't Have to Remember Anything” • Zettelkasten Method
Debunking the myth that search engines and AI eliminate the need for memory. Building deep knowledge through active learning remains essential for critical thinking, despite promises of instant information access.
Howdy, Stranger!
Comments
Aye, look who’s the top story on Hacker News right now.
'“I just ask ChatGPT for that, too!”, the AI generation might ask. Ok, and then what? How can you assess the answers to those three questions (which represent only a fraction of needed knowledge!)?'
Well, asking questions is a good way to learn things, though. To your point about assessing the answers to those questions, it sounds like you are making more of an argument about trusting ChatGPT as a source. I think ChatGPT is actually a great place to start learning about a topic with which you have little experience, but I agree that you should mix in other more traditional sources to have a well-rounded education on it.
Disclaimer: This is the second folder in my favorites in my web browser.
I don't think that is correct for many people. You need very specific knowledge on how to self-learn and, optimally, knowledge on how to learn the topic (which can be delegated to a teacher) to make AI work in that situation.
The problem is that AI can say a lot of right things, while not saying the right thing.
I, for example, never saw any good outcome of somebody learning about health and fitness, if this person wasn't already knowledgeable in this field. But I quite often experience an illusion of having learned well because of AI kissing one's buttock.
Same is true for the Zettelkasten Method. I regularly test AI about its knowledge on the Zettelkasten Method and its quite bad, merely on the level of the run-of-the-mill influencer who heard about it and its publish for his video 2 weeks later.
I am a Zettler
There is no need to remember anything for any process that can perform reliable without human intervention, although you can still do that as a learning exercise or hobby.
However, you need to remember everything involved in analysing and evaluating a concept. You may not need to remember, but you do need to understand. So you cannot really forget anything you might wish to change someday in the future. This trade off acts like a black box that you're interfering with, without understanding its internals.
my first Zettel uid: 202008120915
Going over a post from 2019. Found some remarks that coincide with the point being made in the present piece. Heres’s one:
— Three Layers of Evidence
Emphasis added.
LLMs are cool. DEVONThink’s AI integrations make for a good extension of the standard “Look up...” feature on the Mac. Claude is fun for approximating low-stake abstract thoughts into reasonable associations that can be entertained further on my own. Right now I’m content with this and nothing more.
I think you highlighted the exact technical problem. LLM-content is second hand content.
I am a Zettler