the future of coding

Hey there,

First of all, Happy Tuesday!

Now, if you checked Twitter in the last weeks, you've seen everyone talks about one thing only: Artificial Intelligence.

ChatGPT is getting smarter, and Copilot will soon be able to deploy your application.

If you are a developer writing code for a living, you probably ask yourself if your coding skills are already obsolete and, more importantly:

Do I have a future?

And there are two camps in this conversation.

The first one thinks programming as we know it is dead, that coding will be 100% automated by machines, and that you better look for something else to do.

The other claims that programming is the last job computers will automate. Because programmers do much more than coding, they do creative work and think critically.

Both are wrong.

They are driven by emotion.

The first one by fear (of replacement). The other one by denial.

A more reasonable approach would be first to understand what we are talking about (in this case, Large Language Models, LLMs). And then prepare accordingly.

Why the world will still need coders

AI models are indeed great at generating code. They do it in great numbers, and they do it fast.

But these models are statistical, so they make tons of mistakes. And because they can’t think, they can’t correct themselves. They are by design augmentation tools, not replacements.

Because you still need someone that understands where the AI got things wrong so they can correct it. You still need a Senior Developer.

But Dragos, they are getting better fast. It is only a matter of time before they can do it all... Nope. Not really.

The current architecture based on next token prediction means they cannot reason or “think.” Adding more parameters and data sets won’t fix that.

A complete redesign of how these machines learn would be necessary for AI to operate the way humans do, even for specialized tasks like programming.

They lack the thing we need most in software development.

Why do we have React? Predictable UI state and data flow.

Why do we have Scrum? Predictable output for software teams.

The keyword in the last decades in software development has been... predictability.

Large language models (like ChatGPT and Copilot) increase productivity via code generation but decrease it because of a lack of predictability. They are not reliable. And that’s tricky in an industry where the primary trend has increased reliability.

We are moving towards a future where developers will spend most of their time reading code instead of writing. They will also do a lot more thinking.

What does that mean for you as a developer?

Well, companies will need fewer developers to do the same job. But don’t freak out yet, because as these technologies become mainstream, more and more applications will be built and some hands are needed (still).

Overall generative AI will increase the need for software developers.

Yet, your job as a developer won’t be the same. You will have to transition from the “traditional programmer” focused on the code to becoming more of a guide for these AI-powered tools.

Your best strategy is to get better at “thinking.”

Building mental models is more important than ever. You will also have to learn how these AI models function underneath so you know how to work beside them.

We teach you all that and more in our Software Mastery Mentorship program. And we are back to giving spots after pausing for a while

You can apply here now to see if you qualify! <https://www.theseniordev.com/apply> The truth is no matter how many layers will be between technology and humans; you will always need a more technical specialist. This is and will always be the job of the developer.

Will you have to code less? For sure. Will you have to think more? Undoubtedly. Don’t freak out; keep improving yourself and your skills, and you will do fine.

Because for the next decades or so...

To think is (and still be) human.

Take care, Dragos

P.S. I will start taking calls from Wednesday. Apply here now <https://www.theseniordev.com/apply> and see if we can help you get to the next level so you thrive as a developer in this new Age of AI.

P.S. Wikipedia defines thinking as: “In their most common sense, the terms thought and thinking refer to conscious cognitive processes that can happen independently of sensory stimulation. Their most paradigmatic forms are judging, reasoning, concept formation, problem solving, and deliberation.”