AI Code Generation & The Evolution of Coding
What effect will AI generated code have on software development
A couple of months ago, at the height of Devin fever, I wrote a post stating I didn’t believe you’d lose your job to Devin but that AI was already affecting the job market. Particularly for junior developers.
It was nice to have my opinion validated by the clever folks over at Continuous Delivery 😊 this week.
I have some more thoughts to share on this topic though. Fundamentally, AI code generation is just another step in the journey we’ve been on for a long time making it easier to write code.
When I was a junior developer, if I wanted to use a library I’d have to read the documentation for it. I couldn’t search the internet for an example. I couldn’t ask strangers on Stack Overflow. I’d have to read the documentation and figure it out.
Junior developers today don’t have that skill set (generally speaking). Why read the documentation and figure out what code to write if you can just copy and paste it from Stack Overflow?
Using ChatGPT or GitHub Copilot to know what code to write is simply the next evolutionary step. Sure, you lose something, just as the developer copy and pasting from Stack Overflow lost a little something compared to someone who read and understood the documentation. Thing is, we no longer need every developer to have that level of understanding. I’d love it if we all did, but that’s simply not realistic.
As AI tools evolve, so will the skills developers need to write code. Just as they have so far.
I have yet to see anything though that will replace the function of a human interpreting requirements from another human to turn them into code.
The sheer fact that prompt engineering has become a skill should tell you everything you need to know about how capable AI truly is.
