Overview
I am writing this blog to jot down a couple of pointers to learn properly and deeply in this age of AI. While there are many lists on the internet that address this, I find the following to have worked the best for me as a student and as a programmer. This goes beyond mindlessly consuming content of books, blogs, youtube videos. It is my attempt to streamline resources into a single consumable pipeline which makes it easy to extract knowledge and add value.
Why Are You Learning ?
Learning for most people has become a form of mental masturbation. It is comparable to the dopamine hits that you get from scrolling on your phone except it makes you feel that you have learned something. The reality is that you would forget the intricate details of it by tomorrow and only remember enough to name-drop it during conversations.
Draw the Outline, Build the Project
Often times we choose to learn something new as an exploratory measure without having it connect to any of our existing skills, needs or goals. Learning accelerations come only to those, who have a deeper meaning and clarity behind why they would like to learn something.
People who enjoy learning something outperform those who are innately smarter.
Now the best way to learn something is by building something in the real-world and then hunt for information that is relevant to you as and when you need it. Watching endless tutorials fills your mind with noise and chaos which you cannot seem to bring order about. We have often thrown around the word tutorial hell and that is what this is about. The lack of a project with a deadline allows in all kinds of information. An argument which is often presented is to have sufficient amount of knowledge and pre-requisite before getting started with a project. While this may apply to some very niche domains with a resource constraint, it does not really apply to most others given how we are living in the age of abundance.
Another compelling reason that most people have for being stuck in a tutorial phase is learning to do things the right way the first time around. It is my belief that this thinking of getting it right the first time around and avoiding re-writing and refactoring has often lead to far slower growth and gaining of experience. One ought not to shy away from writing scrappy versions of their project. Something that doesn’t work too well. Something that is held together with duck-tape. Something that does not scale!
When you are learning something new, you must consicously choose to build things that do not cover all general cases but just does the intended job. Your analytical mind would perhaps scream at you for intentionally leaving out inadequacies but you must remember than you are not a child succumbing to the machinations of your mind. It ought to be in your control. Understand the trade-off. Silence the critic and solve the problem in front of your eyes. Build that proof-of-concept and validate your idea before thinking of 10 different problems that do not exist already.
In short, lay your outlines properly! Write them down as 3 goals you have for a project and anything and everything that does not supplement those 3 goals strongly, do not require your attention!
How to Optimize for Pattern Recognition
We recognize patterns only when we are looking for them.
This is often an overlooked aspect of learning but we learn best when we are asking the right questions and are looking answers for them. Often times, we think of our brain as a pool and knowledge as a liquid filling that pool; but in reality a more accurate picturization would be to think of the brain as a mass of LEGO blocks. Every piece of information that is assimilated must fit in this structure but for it to fit, you need to allow it the appropriate grooves that it can grasp on to. When you do not have the right questions, you are missing the grooves and knowledge, like the LEGO blocks that they are, just fall off.
At this stage, something practical and actionable to do would be to break down your tasks into a list of micro-tasks and prompt AI to give you a list of questions that you could find answers to and thereby learn the underlying concept. Bear in mind that the initial set of questions would in no way be the perfect and exhaustive list. But finding answers to them could lead to better questions and help you form the mental grooves to assimilate knowledge.
Pattern recognition is impossible when you are starting from a blank slate. You need something on the plate to feed off of. Allow AI to generate that initial fooder and then put it away as your consume silently before getting back, sharing your learning and progress and hunting for the next set of questions. Make sure to never ask for complete answers directly. Resources and snippets only.
The Fortnight Learner
Learning happens best when there is only one thing to learn. Fragmented focus has always lead to distractions. However, you do need to take breaks, try out different things for a change and sometimes just explore the wild. The important question here would be, how much time to give to something before re-evaluating our perspective on it ?
And that is where the 2-weeks theory steps in. We grossly underestimate how much can be done in a time period of 2 weeks and overestimate how much can be done in 6 months. As human beings, we cannot quantify the velocity of resolving roadblocks while not losing sight of the big picture. Locking ourselves in a 2 week timeline takes away the cognitive load that comes from constantly having to context switch between not losing sight of the big picture while giving sufficient attention to the immediate problem. This is because there is only so much room to veer off on a tangent when you have such a short-time to work on something and build the necessary mental models and patterns for it.
Here, once again, you can leverage AI by asking it to create a 2 week plan in order for you to pick up a stack of things. The prompt ought to invovle asking AI to keep it simple, lean and yet making sure that the density of knowledge consumed is thick. A way to structure your prompt would be to get AI to be the following for you:
- A strategic mentor (initial set of questions)
- A discipline coach (2 weeks only)
- A study regiment (guardrails of how much diversion you can afford)
Deploy In The Wild
The second dose of motivation after the initial spurt comes often when you have things that are broken. But not every broken thing needs to be fixed. Our minds are quite incapable at fixing the broken parts of our project as we often gravitate towards what we find the easiest to fix rather than what is important. And this is where getting real-world feedback on your project is important. We fixate on the is truly important when it becomes a pain point for somebody else.
The broken bits that are truly important are the ones that hurt! As the architect of a project, often times it is easier to be told to fix something and then allow the tunnel vision to re-create your new cycle of 2-weeks.
So we continue. Another conversation with AI. Another list of things to do. Another learning cycle. Overtime, your iteration cycles get faster, the knowledge compunds and you mature alongside your project.
Ritesh Koushik