AI Won't Kill Your AI Won't Kill Your Motivation. But Mediocrity Might.

My last post did something I didn't expect.

I wrote about AI killing my motivation, thinking it might get a few quiet nods from people feeling the same way. Instead, around 180 people read it in a day, some of you followed, and the comments turned into a much better conversation than I expected.

I'm still not entirely sure why. Maybe it hit a nerve. Maybe the timing was right and a lot of people were already sitting with the same discomfort. Could be both. I don't know.

A lot of people came back to the same question: are we still pilots, or are we becoming passengers while the autopilot does the flying?

There was one part of that conversation I didn't really address in the first post. Or maybe I was avoiding it. I'm still not sure which.

For many developers, the uncomfortable feeling is not just about what AI can do. It is about what AI makes harder to hide.

Before tools like ChatGPT and Copilot, it was possible to spend a long time getting by on familiar patterns. Learn React. Learn a framework. Build dashboards. Wire up forms, state, tables, APIs, and permissions.

That work matters. Products need it. But a lot of it is repetitive.

For a long time, being quick at repetitive work could still feel like being highly skilled. There was enough work around, and the difference between "competent" and "exceptional" was not always obvious.

AI changed that. Or it's changing that — I can't tell how fast, honestly.

It can now produce a first pass at the boilerplate, CRUD screens, common component patterns, and code you have written many times before. Not perfectly, but quickly enough that the old advantage starts to feel less valuable.

That part stings. I think that's the core of it.

That can be unsettling, especially if most of your growth has happened inside that category of work. Maybe "unsettling" is too soft. For some people I've talked to, it's been closer to devastating. And I get it.

I do not think that means people are useless or deserve to be replaced. At least I hope it doesn't. That would be a pretty bleak way to look at an entire profession.

I think it means many of us, myself included, discovered that parts of our skill set were narrower than we wanted to admit.

We confused speed with depth. I know I did. I still catch myself doing it.

The developers who seemed least shaken by AI were not necessarily the best at typing code. They were the ones already spending time on harder problems: understanding messy business rules, challenging vague requirements, making trade-offs, debugging production failures, and deciding what not to build.

I've met a few of these developers. They're not always the loudest in the room. Sometimes they're the quietest. They ask one question in a planning meeting and suddenly everyone realizes they've been solving the wrong problem for two days. That kind of thing.

When Copilot arrived, they did not see their whole job disappearing. They saw a way to spend less time on the predictable parts of it.

Tanay made a good point in the comments on the last post. He described AI as an autopilot. It can handle a lot when conditions are stable, but you still need someone who understands the route, recognizes bad weather, and knows what to do when the plan stops working.

That analogy stuck with me. I've turned it over a few times since.

The uncomfortable part is this: if you have only trained for clear weather, AI makes that obvious very quickly.

It did for me. Still does, if I'm being honest.

I sat with that for a while. So what do we actually do about it?

The distinction I keep coming back to is not "AI writes code and humans make decisions." That is too simple. AI can help with decisions too. It can surface options, explain trade-offs, and point out risks I missed. Actually, maybe I'm the one overcomplicating this. Let me try again.

Here's what I keep noticing: give AI a clearly defined problem and it can usually produce a useful starting point. Ask it whether the problem is worth solving, whether your team is accepting the wrong constraint, or whether a shortcut will become painful six months from now, and the answer depends on context it does not truly own.

It does not sit in the meeting when a stakeholder changes the goal halfway through the conversation. It does not carry the consequences when a technical decision creates maintenance work for the rest of the team. It does not know which compromise is acceptable for your product, your users, or your company.

That responsibility is still ours. For now, at least. I don't know what happens in five years, and anyone who tells you they do is guessing.

The developers I see doing well are not simply writing less code. They are getting better at framing problems before code exists.

They read more code. They ask better questions in planning. They clarify requirements that sound obvious but are not. They learn enough about the business to recognize when a technically clean solution solves the wrong problem.

That is not glamorous work. A good decision in a meeting does not show up as a polished screenshot or an impressive GitHub diff.

But it changes what gets built. I think. I'm still figuring out how to measure that.

I am still figuring this out myself. Some days I worry that I am not learning fast enough. Other days I look at what I shipped last month and realize that, without AI, it would have taken a much larger chunk of my time.

Maybe that's the whole thing right there. Not getting faster at the old work. Getting better at the new work that the old work was taking time away from.

The goal is not to compete with AI at the tasks it was designed to accelerate.

The goal is to use that time to become better at the parts of the job that need judgment: understanding people, understanding systems, spotting bad assumptions, and taking responsibility for the outcome.

AI did not kill my motivation.

It made me look more honestly at where my motivation came from, what I was actually good at, and what I still need to learn.

That part is uncomfortable. But it is also useful. At least I keep telling myself that.

---

I'm a frontend developer based in China, building Dual-Track Dev — a blog about Chinese developer tools and trends for a global audience. If you're navigating the same AI anxiety, reach out or drop a comment. I'd love to hear how you're handling it.