When GPT-4 dropped, then Claude, then all these coding assistants that keep getting better every month, I thought I’d just be excited. I mean this meant faster development, fewer stupid bugs, more time to focus on the interesting problems. And yeah, all of that happened.


But something else came with it.. A weird sense of demotivation that I can’t shake off.

They are Becoming Too Good

Let’s be real. These AI tools are insanely good now. They write boilerplate in seconds amd they even refactor without complaining in no time. They explain confusing code better than most documentation would ever do, and they generate tests, configs, edge cases, all the stuff that used to eat up take up so much time and mental energy.


And they’re getting faster, cheaper, and more accessible with every new model. Every week there’s some new model that’s better at coding than the last one. Its going way too fast.


Which is incredible. Like genuinely amazing if you step back and think about it.


But a lot of the satisfaction in coding used to come from the struggle. Figuring something out after being stuck for an hour. Finally getting that one piece of code to work after digging through documentation, StackOverflow, and trial and error. Slowly building something that only existed in your head into actual working software. It was just way too rewarding and satisfying.


Now a lot of that satisfaction is just gone. You describe what you want, the AI writes it, you tweak it a bit, and you’re done. Five minutes instead of five hours.


Yea sure like, I could just stop using AI models to code. But when I think of the value they add, it’s hard to justify that. It’s like saying “I want to do this thing the hard way because it feels more rewarding.” Which is a valid choice, but it doesn’t change the fact that the tools are objectively better at certain parts of the job than I am.


And since time is money, and I want to be efficient, it’s hard to ignore the benefits. But at the same time, it feels like I’m losing something important in the process.

The Work Doesn’t Feel as Valuable

There’s this weird psychological thing that happens when difficult tasks become easy. Your brain quietly adjusts how much it values the output. If something takes five minutes instead of five hours, it feels cheaper, even when the result is better, hence the satisfaction drops.


This isn’t even about ego, or at least I don’t think it is. It’s about the meaning of it.


When an AI can produce code that looks pretty similar to what you would’ve written after thinking through the problem yourself, it’s hard not to ask: what exactly is my role here?


Not being useless, I’m not saying that. But just different, maybe less obvious, and less clear?

The Job Market Is CLEARLY Shifting

There’s so much noise about this right now. Some people say developers are completely doomed. Others insist nothing will fundamentally change. Most of us are somewhere in the middle, quietly anxious.


But just seeing the historical low developer/programmer job openings, it is at least very clear to me, and i’m sure to many others as well, that the shift is already happening. Companies are hiring fewer developers than they used to. The demand is still there, but it’s not growing like it was before.


It’s not that AI is replacing developers entirely. It’s that it’s replacing parts of what developers do. And when those parts disappear, the need for as many developers goes down.


And when significant parts of a job disappear, you need fewer people to handle what’s left. That’s just basic math.

The Experts 🧢

AI is really good at pattern completion. It’s seen millions of code examples and can mix them in clever ways.


But it’s terrible at responsibility.


It doesn’t understand context beyond what you explicitly tell it (but its getting there). It doesn’t own the consequences when something breaks. It can’t sit in meetings and negotiate tradeoffs. It can’t deal with ambiguous requirements or figure out what the stakeholder actually wants when they don’t know themselves. It can’t make judgment calls when the specs are wrong, incomplete, or politically motivated.


Real software work isn’t just about writing code. It’s really about understanding the people, systems, constraints, risk and timing. It’s knowing when to say no, when to push back and when to build the quick version versus a more robust version.


So like even though AI can help with the execution, humans still need to define the intent, make the judgment calls, and take responsibility for the outcomes. And that’s not something AI can do. At least not yet…

Why It’s Still Replacing Some of Us Anyway

A lot of developer jobs existed mainly to move code from software specification to implementation. Junior roles writing CRUD apps, contractors building internal tools, teams maintaining repetitive services and people writing glue code between systems.


If one senior developer with AI can now do the work that used to require three people, companies will absolutely notice. Not necessarily because they’re evil, but because they’re businesses optimizing for efficiency. So you can’t blame them for wanting to get the same amount of work done with fewer resources really.


But that’s also what is creating the pressure on the job market.

The Motivational Problem

Before jobs actually disappear, motivation takes the first hit.


People stop feeling “special”. Stop feeling proud of their work, and stop feeling like the skill they spent years building is actually rare or valuable.


Coding starts feeling less like a craft and more like orchestration. You’re not the one carefully typing every line anymore. You’re the one deciding what to build, what the AI’s output is good enough for, what needs to be rewritten, what to trust and what to throw away.


But this just makes the work harder to say “I built this.” It’s more like “I directed this.” Or “I curated this.” Which is still valuable, but it doesn’t have the same emotional payoff as writing every line yourself.

The Shift Nobody Prepared Us For

For years we were told to learn to code, that it was the safe career path and the future-proof valuable skill.


Then the future showed up earlier than expected and said “actually, I can handle most of this now, thanks.”


That does something to people. Especially those of us who genuinely loved coding for its own sake, not just as a paycheck.

The Awkward Middle Phase

Maybe the demotivation isn’t because coding has become pointless. Maybe it’s because we’re stuck in this awkward transition period.


I don’t see myself or other developers as obsolete, and I don’t think coding is dead.


It is just the old rewards that are disappearing, but the new ones aren’t clear yet. The tools got better way faster than our sense of purpose adapted.


We’re all trying to figure out what our role is supposed to be now. And that gap between what we used to do and what we’re supposed to do next feels uncomfortable and uncertain.


But uncomfortable doesn’t mean the work doesn’t matter anymore. It just means the reason it matters is changing, and we’re still catching up to that reality.


I don’t have a neat conclusion here. I’m still figuring this out myself. Some days I feel energized by what AI enables. Other days I wonder if I’m just training my replacement.


Both can be true at the same time. And maybe that’s just where we are right now?

Update on This one Year later

Yea fuck this.