How to Self Study Technical Things in 2025

So I've been learning hard technical things since the age of eight when I started programming and I think after studying physics at Cambridge University I become even better at it and these are basically all the things that I wish I knew when I first started and how you would learn technical things now especially with the help of AI. So

number one is to use project based learning. I think a misconception or trap that many people fall into is that when they want to learn something new such as Python then they go on YouTube and they search Python and then they find a 20-hour long video by free code camp that supposedly will teach them everything they need to know. and they think the longer the video they watch about like Python, the better it will be for them. And about one two hours into a video or course, they get really bored and then they just give up because it's just too boring to watch the entire course and there's not much to follow along and do with the course. I think generally a better approach is that you want to find the shortest video that you can. It could be a 1 or 2 hour long video about Python that teaches you most of the basics and then you would find projects to work on that will help you improve your skills. The point in project based learning is that when you're working on a project, you will encounter problems and you won't know something and then you'll have to go find and learn the thing. And that thing that you end up finding and learning will stick in your head for much longer and be much better conceptually for you because you knew the problem it was trying to solve. I think when you're watching a 20our long course, you're presented with 20 30 plus solutions for problems that you don't even know like exist. And because of that, since you have no problem that you've experienced, it's much harder to remember the solution for it. So you would likely have to make a significantly more notes if you're not doing project- based learning. And whatever you do learn, you'll likely forget much sooner. Ultimately, I think you can't fully appreciate technical solutions unless you worked on a project in which a problem arose that required that solution to begin with. So yeah, with project- based learning, I think you should ask Chad GPT something like based on what you know about me and my interests, what projects would you recommend to help me learn X? And that X could be like Python, JavaScript, databases, or just anything else you have in mind. So ultimately the cycle becomes that you learn enough, you try doing some projects, you run into problems, you figure out what knowledge you're missing and then you go and learn that knowledge and then you continue working on the project. And I think this solves a problem that a lot of people have where they're just reading a book or they're watching a really long course and they're not following along or doing any projects and then they feel burnt out because of that. And it's because it just becomes very boring. Like you spent 20 hours watching a course and you don't have anything to really show for it. Whereas if you spend 2 hours watching a course and then you work on 10 projects for the remaining 18 hours, then you have a lot to show for it and you feel more motivated with each project that

you're working on. So the next thing I do is kind of based on the fineman technique and basically Richard Fineman was a Nobel Prize winning physicist. And what he used to do is whenever he was learning something new, then he would learn as much as he needed to and then try to explain that concept to someone else. And then in the process of explaining, he would realize there's something he doesn't quite understand or there's some kind of deeper thing going on that he can't quite explain. So he would go back and learn that and then he would come back and explain it in more detail than before. And basically in the process of explaining something someone else or teaching someone else, you identify the gaps that you have in your own knowledge and then you can also cement your own understanding of whatever concept you're explaining. So what I personally like to do is when I'm learning anything technical, I try to rephrase or rewrite what I just understood and explain it to Chad GPT. And then basically I get Chad GPT to correct my explanation or give me a different way of thinking about it. And I think it's pretty good when you're like using AI or Chad GPT or someone else because even on Chad GPT voice mode you can try and explain something to it and if you're wrong you can get it to

correct you. Now the next thing is to use AI to help you write simulations. Something that you might notice when you meet with some really smart people is that they have a sense or like intuitive feel for what happens if you change certain conditions. So that could be especially in physics they would kind of know without working through the maths or the equations what would likely happen if uh you introduce another variable or factor or something else. And that's because they kind of have a feel for the concept. So they have this kind of like loose feeling for the equations and so forth. It's kind of like if you work between many different currencies and you convert money frequently, then someone can say a number to you and without you having to do any math in your head, you can kind of guess roughly what that would be of a currency. So for example, one thing you could do is you could use cloud 3. 7 sonet to write code for gravitational lensing around a black hole, which this person did, and then have it give you parameters that you can like adjust. So you can get a feeling for what happens in each situation when you adjust different parameters and I think that should give you a better understanding of certain concepts if it applies.

Another thing related to projects is to copy something that someone else has done. So often when people are learning machine learning over ideas, what they do is they find a research paper of a concept they particularly like or they want to implement themselves and then they follow the techniques outlined in the research paper to see if they can get the same results and during that process they learn a whole lot about the research itself and they develop understanding a lot of different areas and this also happens when people are making mobile apps for example. So, one of the first apps that many people make when they learn to make mobile apps is a habit tracker app. And that's because they probably use a habit tracker app themselves and they have a lot of motivation to make a better one that's nicer for themselves as well. So, ultimately, if there's something that you already use yourself, you might want to make it again yourself because you will have more motivation to make it. There are people who learn game development and they try to make their own version of Minecraft. And even though they only get like 5% of the way there, they still learn a whole lot during that process that is transferable knowledge to other domains and potentially any other games that they would make. During the earlier stages of learning anything new and technical, you're going to feel like you constantly suck and should give up for something. And that can be especially hard if you're working on a project that you're not motivated by because you watched a bad in a course. And that could be, for example, you're learning app development and you're making an alarm clock app and you don't care about that and you'd rather be making an app like a habit tracker app or just like a video game instead. It's better to work on a harder project that you're more motivated by because you're more likely to reach your goal at the end of whatever you want to learn and make than it is to work on an easier project that it was just shown in a course or video or tutorial for you. The

next thing to do is to set a reward cycle. And reward cycles are basically everywhere. So, for example, if you're playing a video game, then you might be battling a dragon, you get coins, then you take damage and then you like buy new gear and then you level up and you move on to next level. Or it could be you're learning a new recipe because you really enjoy cooking and then you make the food and then you get to try it and it's really tasty and then you feel rewarded. Basically, what I noticed for myself is that I really enjoy activities where the time input and the time output or the delay between the two is really short. So, for example, I really enjoy app and web development because when I make changes to like a few lines of code, I can press save, then refresh a page, and then I can see if I'm right or wrong. If I'm wrong, it would give me an error. If I'm right, it would do what it wants, like I want it to do. And I didn't enjoy, for example, like physics labs because I would get the result back like a day or two later in some cases and I couldn't iterate quite as quickly. So if you're in that group or category where you feel like you want short feedback loops and you want to iterate quite quickly, I would suggest learning and focusing on technical things that are better suited towards that. So for example, like if you're making machine learning models and you have to wait 3 days for a machine learning model to train and you feel like pretty unmotivated during the time in between training, it might be better to work on something else entirely that you feel more motivated by and you can get feedback on sooner. So for example, I also enjoy making videos because I can usually make a video and then an hour or two later I can upload the video and then I can see feedback and get comments and so forth from the video. So basically if you really enjoy short feedback loops and I would recommend finding a way to have those short feedback loops in whatever you're trying

to learn. And finally is accountability. A common problem that you might have with self-studying is that you feel like you're not held accountable in any way. So for example, when you're in college, you have assignments due, you have a professor asking you to hand in work or something, you have tests, you have classmates or class projects to work on. Basically, there are many things holding you accountable to actually getting the work done. Whereas, if you're self-studying, there's likely nothing that really goes wrong if you don't learn what you were set out to do this week, and that can be problematic. So, you have to find external sources of accountability yourself. And that could be like if you're making a product for other people, then you want to get initial users for your product as soon as possible because as soon as you have users, whether that could be for an app or a software or a service or something, you will feel a desire to constantly improve the product for them. and you might be getting feature requests and support tickets and so forth. So, you would feel motivated to continue learning and improving. You can also try and find an online community for whatever you're trying to learn and that can be on Discord or Reddit or some other place. But you want to make sure you find a community where there are not many toxic people or zero toxic people. Because often what happens is when you're on these communities, there'll be one or two or a couple of people who are really like mean and they will discourage you from learning and they'll like give you bad advice or wrong advice that they know is bad advice as well. and that can be really harmful in the long run. So, I would recommend finding a nicer community which is well managed as well and they kick out these toxic people and I'm actually planning on starting my own community especially when it comes to building things with AI and that will be linked down below in the description if you want to if you're interested in signing up there's a Google form for it. And the next thing you might also want to consider is live streaming yourself. So, one of my friends from MIT whenever he was working or studying, he would live stream himself working and studying. And even if there were one or two people watching on Twitch, he would just still feel motivated because it feels like someone is watching over him. The reason why libraries work for a lot of people is because when you're sitting in a library working, it feels like people are watching you and they're judging you if you're not working. Whereas, if you're working alone in your room, it feels like no one is judging you if you basically don't work. Another way of accountability can just be spending a small amount of money on something. When you spend money on a course or a book or anything else that you're learning, then you to avoid certain cost policy or to feel like you wasted your money will actually complete what you're set out to do. If you compare the prices or like the completion rates of paid courses versus free courses, you see that paid courses generally have high completion rates because people are more invested from the beginning of actually accomplishing that goal. It's kind of like going to a gym and then saying to your friend that if you don't go to gym once like every week, then you will basically give them $10. But yeah, I think all those different forms of accountability will lead you to form better habits. And those habits will keep you going in the longer term when it comes to learning anything new and technical than all the accountability measures themselves. But yeah, like I

said earlier in the video, I'm thinking of making my own community for people who are serious about making things with AI and that will be well moderated and stuff. And if you're interested in joining, then fill out the form down below.