We have moved from actively writing code by hand, as this was arguably the fun thing to do in programming: the feeling that we were crafting things from scratch and watching it power engines and support communities around the globe.
However, in this transition where code is cheap, the goalpost has shifted to actively testing what was built by your AI agents: quality assurance, translating users' problems into well-defined tickets, more time spent on reviewing code. Most of these are things the average engineer dislikes.
Writing code by hand isn't going away entirely, as we still need engineers who understand code to debug and, importantly, architect systems, and you can only architect what you know how to build. This craft isn't dead, but evolving.
We now need to ensure that what we are building with our agents is reliable, maintainable even for human readers, and that performance is not degraded. I must say, with AI we should be able to do many of the checks and ensure things are in the right places.
The developers who only identify with code will struggle to adapt. Skills that matter are shifting, figure them out and learn them.
Top comments (0)