AI Era’s Unconventional Learning Method: How I Went from a Beginner to a Development Expert

No tutorials or books needed, master a new language in 1 day (not for 99, not for 9.9, just if you’re willing to use it – classic sales pitch). This isn’t a dream, and it’s not an ad! AI models make learning a new language in no time possible.

Of course, this statement is a bit exaggerated. What I mean is that in the AI era, we no longer need to learn a new language the traditional way—learning first and then practicing. Instead, we can learn directly through practice.

This is my reflection after using Cursor for one month to learn front-end development.

I’m a front-end beginner. After a month of secondary development on an open-source project, I not only “entered” NextJS development but also started a new passive income project. I’m now exploring the advanced features of NextJS. Why the quotation marks around “entered”? Because I didn’t enter in the traditional sense. I didn’t master JavaScript or TypeScript syntax. I simply learned how to use AI for development. To put it simply, I can develop but I can’t write code.

Front-end development is an essential skill for my freelance career. I happened to see a front-end open-source project on X, so I decided to use Cursor to work on it and learn front-end development through practice.

My front-end knowledge was still stuck in the CSS, HTML, and jQuery days over a decade ago. I didn’t even know JavaScript, NodeJS, or TypeScript.

Normally, to learn front-end development systematically, you would start with HTML, CSS, and JavaScript, and then move on to learning NodeJS and TypeScript.

But in the AI era, I didn’t follow the traditional learning route. Instead, I rolled up my sleeves and started coding. First, I learned from the existing project’s code, and then deepened my understanding through secondary development.

1 My Experience of Learning a New Language with AI

1.1 1. Letting AI Develop Tasks Independently Without Knowing Code

At first, I followed the code AI gave me completely, but the result was that the code didn’t run at all. This is because current AI models, without functionality and design documentation, can’t accurately understand existing code, let alone independently complete development tasks. It would write code based on its understanding of part of the code, but it would be complete nonsense, and it couldn’t complete the secondary development of the existing project.

1.2 2. AI Becomes the Consultant, and I Become the Commander

No choice, the AI models aren’t that smart yet, so I had to treat it as a consultant. I would ask it for execution suggestions, and then research and confirm my own development plan.

I’d follow the steps in the documentation, let AI execute them, and then I’d check the code and ask why it was written that way. If there was code I didn’t understand, I’d have it explain it to me.

As I read more documentation and got more familiar with the code, I eventually understood how to develop it, but I still couldn’t write the code (I didn’t know which basic functions and libraries to use), so I had to ask AI to write the code for me.

1.3 3. Optimizing AI-Generated Code

Once the functionality was implemented and the program ran, it was still a bit slow. So, I did more research and slowly optimized the existing program. This is when my NextJS skills went from beginner to advanced.

2 Drawbacks

At the moment, you can’t fully trust the code AI generates. If you ask it to fix a small issue, it might end up tearing down the whole road and repaving it, even digging up the fiber optics. The solution AI provides is just one of many options—it’s not necessarily the best one.

From an engineering perspective, the code AI writes cannot guarantee reusability, scalability, readability, best practices, or consistent coding style.

I often run into X-Y problems, and AI isn’t intelligent enough yet to identify these kinds of issues. For example, when a file doesn’t exist, the code throws an error. But I tried handling the error using the Golang method, which doesn’t suit TypeScript. The correct way would be to have AI handle the “file doesn’t exist” case rather than directly handling the error.

3 Takeaways

1. The market for basic tutorials and beginner books is shrinking. Because the experience of learning with AI is much better—it’s like having a 24/7 private tutor who can answer questions anytime and even help you write code.
2. There’s less need for junior developers.
3. For areas requiring strong experience accumulation, AI still can’t replace it. For example, solving system bugs, optimizing code, understanding best practices, architecture design, etc.
4. Senior engineers + AI = productivity doubled, it’s like adding wings to a tiger.
5. The one-person company or super-individual model will become more common.
6. The ability to quickly master a new language is thanks to the maturity of frameworks and the power of AI models.
7. Learning has shifted from “learn first, practice later” to “learn by doing, with problems to solve.”

These methods are probably better suited for those with some computer science background.

Finally, here’s a link to my practice project: a tool for generating social images and YouTube banners.