AI-based learning is a joke

2026.09.10



Initially I was looking for some resources for mastering x86 ASM and this is when the motivation to write this post was born. After asking a bunch of people I had nothing but a just use AI to learn x recommendation. Don't consider it personal.



Learning is fun. Always has been and always will be. However, in the modern era of Artificial Inteligence (LLMs for now) people don't actually learn anymore. They prompt and immediately get the answer. There is no learning here, just some wording, and copy-pasting the answer. And even if you don't use any external help from this black boxes you can have this thought at the back of your head to maybe prompt and get an immediate answer.


This video1 explains the whole concept of being a beginner better than I ever could but I got some thoughts myself.


I don't consider AI bad by nature. It is a good technology that just got capitalised by a group of evil people. Trillion dollar companies want to your data no matter the cost2 and your personal privacy is up to you, and only you!


When you start learning a new thing you are a beginner and you will not know all small details from the very start. And that is okay. Being a beginner means to research, to look through a ton of information that you will probably remember. This is how the learning process and the brain work.
The problem is that most of the people don't know how to use LLMs properly. They start with a confident prompt and eventually after some time the whole magic of "explain like I'm five" ends.


Before AI era developers had to prove their knowledge on x-topic, their worth and value. And having a degree meant a lot more back then than it is now. As a developer you either knew the material or you didn't. All problems were on you. If you didn't that x, well, that meant getting some books, reading StackOverflow, asking on Reddit. These days knowledge can be easily compromised.


To not sound like some schizo, AI is good at a lot of things. For example: automating something generic, at doing stuff that you probably wouldn't be bothered doing, something else.
But what's the point of automation if every task should be supervised and controlled?


There is also another problem that Silicon Valley created.

I wouldn't consider it a problem but rather a development in requirements. In a modern world a Programmer should know almost everything related to a field and beyond it, have 100500 projects, 5+ years of experience and all of this is just for some Junior Dev position that pays ~25k euro a year. Employers search someone who is a Jack of all trades, master of all.. Specialization is basically dead.
And if you don't use AI to learn, or to ship projects you will be left behind. Or so they say. At least you'll get this feeling after being constantly on LinkedIn.


Personally, I don't use AI. I like to learn by myself. I'd rather suffer through the process than to just get some generated answer. This is how one learns.


Don't ever trust AI. Unless it is local.


If you have any comments, please email me: batistaq [at] proton [dot] me




References


  1. How to Become an Expert at Programming
  2. Claude Shared Chats Surface in Search Results Containing API Keys and Personal Data