The more you use AI, the less technical you become.
AI will probably make you more productive. You finish tasks faster, close more tickets, and produce more code. At the same time, it's making it easier to stop doing the thinking that built your technical understanding.
You can get more done and understand less of what you've done.
That's the part of the productivity conversation we keep overlooking.
The steps you're skipping were teaching you
Before AI, an error meant you had to investigate. You read the message, found the relevant code, checked the inputs, and traced how the data moved through the application.
Sometimes it took hours. Along the way, you learned how the system worked. You discovered assumptions you hadn't noticed and dependencies you didn't know existed.
Now you can paste the error into an assistant and ask it to fix the problem. It changes the code, the error disappears, and you move on.
The task is complete. The understanding you would have gained from solving it never develops.
Repeat that often enough and you become dependent on AI for problems you used to work through yourself. Your output goes up while your critical thinking gets less practice.
The tests pass. Nobody asks why.
A lot of developers and engineers have stopped reading error messages properly. They don't trace how the data flows. They don't examine how the code was fixed or ask why the change works.
If it passes a set of tests, they accept it.
Then the review goes to another model. An LLM judge examines the changes and gives its verdict. The AI wrote the patch, the AI reviewed it, and the human approved the result.
There may be useful checks in that process. But where did you have to understand the problem? Where did you question an assumption or decide whether the implementation made sense?
A successful test run tells you that the checks passed. It doesn't mean you've understood the code, and it doesn't tell you what nobody thought to test.
Your stakeholder has the same tools
While engineers are handing more of their thinking to AI, non-technical people are using it to do work they previously needed an engineer for.
A stakeholder who once waited for you to build a dashboard can now ask Claude to generate one. They can change the layout, add a chart, and connect a data source without putting a request in your queue.
They will still encounter problems. But look at how both of you respond: you paste an error into AI, accept the fix, and check whether the result looks right.
If that has become your entire workflow too, the difference between you is getting harder to see.
The gap between technical and non-technical people is shrinking. Some of that comes from what AI lets them do. Some of it comes from the technical work you've stopped doing yourself.
What are you still learning?
You can spend a whole day producing code without learning much about the application you're changing.
That should concern you more than how many tasks you completed.
Read the error before asking for a fix. Follow the data through the code. Understand the patch before accepting it. Ask why it works, then check the answer against the implementation.
Those habits take time. They are also how you keep developing the judgement your job requires.
AI is making it easier for anyone to produce a working result. If you keep outsourcing the steps that make you technical, what will distinguish your work from theirs?
Comments