Overcoming AI anxiety

Writing code pays my bills. It also happens to be a hobby that I was fortunate enough to turn into a career. The satisfaction of solving a problem, the concept of building something from nothing, the small wins that add up over time, the sense of accomplishment. That's what I enjoy.

For the longest time that felt like a stable foundation. If you could write good code and build solid software, there would always be a place for you somewhere.

Enter AI.

At first it was tab completions. Little suggestions inside of my editor that gave me a little productivity boost, not too different from a language server providing me with smart and sensible completions for variables and types. Easy to accept with a Tab, easy to ignore with an Esc. It felt incremental.

Then it stopped feeling incremental. Those small suggestions turned into entire functions, entire classes. You soon realise you're not writing every line of code and you're no longer engaging with your code in the same way. You move from construction to curation, spending more time reading, steering and second-guessing.

That's where my unease started. Unease is probably a poor choice of word because the tooling is reliable, often remarkably capable, and almost as good as you (sometimes better to be honest).

My own "anxiety" around AI wasn’t really fear of the technology itself. It was the realisation that something fundamental about the way I work was shifting. I wasn't building things up piece-by-piece in the same way anymore. I was merely an evaluation engine and guidance counsellor for a "clanker".

Once you sit with that for a while, you start to ask the hard questions.

"If this thing can write most of the code, what value do I add?"

"If I'm not the one writing the code for every solution, will I lose touch?"

And perhaps more quitely, "If this keeps improving at the same rate, where does that leave me in X years?".

I didn't really have answers for those questions immediately and that's where the "anxiety" really lingers.

It was only once I had truly integrated these tools into my day-to-day work that I started to see the obvious boundaries and form real answers.

AI generates code, but it doesn't truly understand the intent. It doesn’t always have the context to take a real problem in your system and map it to an outcome. You're the one who bridges that gap. It follows patterns and guidelines that you provide, but it doesn’t own the consequences of those choices. It can suggest a solution, but only to the problem you’ve framed.

That distinction is what matters. The responsibility of software design and software engineering hasn't moved. You're still the one deciding what to build, why it matters, and whether the solution is actually correct. Not just syntactically, but architecturally and contextually.

The AI accelerates your execution, it doesn't replace your brain and your judgment. Once that clicks, those questions start to feel a little less open-ended and existential.

To leave you with a more poetic outro:

Your value as a software engineer isn't in writing every single line of code, it's knowing which lines should exist at all.