Or well what is the riskiest thing that they can cost us, understanding. And understanding at various levels not just of the code we are writing but of the problem space we are working on. Sure if you are writing yet another logger then auto generate that thing, assuming you have done this a few times and the problem space of a logger is not new to you.
A long time ago like most things in computer science some dude made the argument that one of the most important artifacts that comes out of the programming process is a shared understanding of what the problem is and how this program address. Mostly in the context of the life cycle of software but I will argue that it also applies here.
Aside from that I will argue that it is still important to do the boring repetitive things specially early on in your career, they might seem like a waste of time but being able to build up that experience overtime is what will help you move quicker.
To illustrate, what is the point of learning the the times tables in school? Its just route memorization to an extent. But not being able to do them quickly does hinder you later on in once you get to more advanced subjects since its important to remember that you are limited in the amount of things you can hold in your head. The number is 4 chunks at a given time. Say you are learning some algebra you if you took the time to get good at your times tables or arithmetic in general that's one less slot that it has to occupy as you try to learn whatever you are learning. Otherwise you end up having to hold two things in memory at a given time and that is hard.
The same applies here to learning how to program. I can read good code faster than I can write good code (however I can write bad code really fast) so that allows me to generate code and quickly decide, yep this is good, or nah this is awful try again. But I can only do this because I have done it by hand over and over.
Anyways that is hard because in general its going to be an opposing view to what a business is going to want out of you when it comes to making a product. Never mind that it will hinder them in the long run due to the lack of understanding we talked about earlier. Solving this issues is a bit out of the scope of a talk aimed for juniors but it is still something that is fun to think about.
The biggest thing LLMs could cost you isn’t your job. It's your understanding. If you rely on AI to think for you, you stay stuck forever needing crutches. If you push through the pain of real learning, you move faster later. It's just like learning your times tables as a kid — you hate it then, but you thank yourself later.
Top comments (0)