AI’s Biggest Threat: Young People Who Can’t Think - WSJ

This is a great read, and it’s so true.

I’ve experienced this myself as a CS student over the past year. I lost a lot of the programming skills that I had and also didn’t really learn a lot of new things even though I built a lot of new projects. The reason is that I offloaded a lot of things to AI.

I didn’t really do any programming myself over the past year. It’s too tempting to give the task to AI because it will get it right and it will be fast.

The result was that I built a lot of cool projects. It definitely felt very productive. But, I didn’t know how they worked, and I didn’t learn any skill from them.

Now I’m sitting here at the end of the year, realizing if I had put in the time and effort myself, I would have such a bigger skillset. So, the past year was really a waste of time and opportunity to learn.

Similarly, I used AI to write most of the reports I had to write. In fact, I think there wasn’t one piece of work where AI hadn’t generated 90% of it. Now, when I want to write something as simple as a blog, I find it hard to structure what I want to say.

So, the harsh reality is that I’ve lost most of the skills I had, and haven’t really learned any new skills because of improper use of AI. Improper not in the sense of plagiarism or illegal, but in the sense of being irresponsible to my own long-term interests of learning new skills.

I’ve been trying to change this, though. I’m trying to learn back those skills, and learn new skills, with the occasional help of AI but seriously avoiding offloading the tasks to AI.

Learning comes from doing, getting stuck, doing in a different way, realizing it was wrong again, and doing it again. This is how you crystalize the concepts in your brain and learn the connections between them. This is impossible with AI.

So, I’m giving myself the time needed to learn properly, because it will take more time than if I just offloaded it to AI, but I will be deprived of learning any skill.

To do this, it’s crucial to distinguish between learning vs doing as your goal. If you’ve learned something in the past and have already done it many times, and now you just want to be fast, use AI to the best of your abilities. You already have the skill and experience to guide the AI and realize if it makes any mistakes.

If you’re learning something new, however, you should limit the use of AI to a slightly more powerful search engine. You should be the one typing on the keyboard. At some point, you might realize it was wrong, so you can start again. Ask AI how you could prevent the thing from happening again, but rebuild it yourself.