It was never about writing code.
A lot of us coders became coders because it looked and felt cool to write code. Seeing your first real programs run made you feel like you had created something alive!
We all have our pet projects and hobbies but in our daily jobs we are creating applications for real-life use. In that world we have always been between the customer or business needs and the computer. The meat proxy between dreams and machines.
The meat proxy.
Before those dreams are actualized as a product, people have to make design decisions in software architecture and visual design. This part of the process hopefully never disappears, because when we give the AI more control and disconnect ourselves from the systems, we paint ourselves into a corner.
When the part of producing the code is removed from this process, it doesn't mean that we don't need people anymore. In a way we all got promoted to a product manager who knows how to code.
AI-slop or AI-gold?
We should look at using coding agents as if we were doing pair programming. Actually read and understand what it produces and push back on its decisions sometimes. This also means that we won't ask for too much at once! You are the one making the plans and decisions, not the AI, that's the most important thing in the process.
What is AI-slop and is everything AI produces considered slop? No. Slop happens when people get lazy or don't understand what they are doing or requesting from the AI. When you don't review the results and don't push back and demand better results. And you can't demand better results or coach the AI to do better if you don't understand what you're doing.
Sometimes we call buggy messes AI-slop, like we didn't ever write bugs ourselves...
Make the AI agents do the tedious tasks you know how to do but never cared about starting: Analyzing data sets, doing code refactors, etc. Start thinking about the bigger picture and make the coding agents work in small phases that you can easily oversee.
Conclusion
Software engineering was never about just writing code. Will leaning on AI too much be bad for us? In many ways, yes. In case the AI bubble bursts, whether it be from ever-increasing pricing or the impossible amount of resources it demands, we shouldn't forget the craft of coding, nor rely on AI too much.