There is a lot of talk about how AI is going to kill software engineering. I wanted to put my two cents on the matter.

AI will not kill software engineering

Anyone can code. It’s a skill, and unlike most skills, it’s incredibly accessible. You just need a functioning laptop and internet access, and you are ready to go.

When you are a professional, it’s not just about coding. It’s about learning to manage requirements and business needs so you can make progress even when things are not clear-cut.

In the past, if I was doing greenfield work, like at a startup, and I needed boilerplate code or scaffolding, I’d throw myself into 12–14 hour workdays for a little bit just to get the foundation work done and then chill, focusing on requirement gathering and evolution. Essentially treating the project like a block-in and pivoting wherever I needed to. To be honest, this is the kind of work AI has thankfully killed. My lower back is very thankful for AI.

However, software engineering is about the evolution of the project. Sometimes that’s small changes; sometimes it’s a rewrite. But evolution always comes down to asking the right questions.

That’s hard. You still need training to do that. You need to learn to question every requirement — and to not chase every requirement.

There is an art to building software when the requirements are always changing. There is an art to building software when you have to support 10M financial transactions monthly. You need to learn to see the patterns in the answers to the questions.

Engineering has always been about the questions you ask and the constraints you operate under. Anything else is trivial.