Why 'AI Will Replace Coders' Is the Wrong Conversation

There's a version of this debate playing out constantly on X and LinkedIn right now. Someone posts that developers who lean on AI too much are finished. Someone else replies that token usage will decide who gets laid off next. Threads pile up. Everyone sounds very certain.

I think most of it is noise.

The Real Skill Was Never Typing Code#

Here's something I've come to believe after watching this play out on my own team. Writing code was never the hard part. Anyone who's spent real time in this field knows that the actual difficulty was always in the thinking — breaking a vague problem into something solvable, figuring out what could go wrong, deciding which tradeoff actually matters here versus which one just sounds important.

AI didn't remove that requirement. It just made it optional to engage with, in the short term. And a lot of people took that option.

What I've noticed is that the people getting nervous about “AI replacing developers” are often the same people who quietly stopped doing the FAFO part of learning. You know the one — where you try something, it breaks, you sit with the breaking for a while, you figure out why, and that friction is what actually builds the skill. AI can shortcut the breaking part. But if you let it shortcut the sitting-with-it part too, you don't end up faster. You end up without the skill.

What I Got Right, By Accident#

I got lucky in one specific way. My senior at work, Purnachand, never stopped me from breaking things. He let me FAFO when I needed to, and at the same time, he was the one actively pushing me to figure out how to use AI well — not as a replacement for thinking, but as something that sits next to it. That combination mattered more than either piece alone would have.

I've also had a similar experience working with Ayush Chugh, who's been genuinely great at simplifying things — taking something that feels complicated and showing you the actual shape of it underneath. That kind of collaboration is exactly the thing missing from a lot of these online debates. The conversations where someone helps you see your own problem more clearly don't show up in a prompt history. They show up in how you think six months later.

My Honest Relationship With AI Right Now#

I'll be honest about where I actually am with this, because I think pretending otherwise doesn't help anyone, including me.

Most of the time, I try to use AI as a thinking partner. I bring the problem, I bring my own half formed opinion about the approach, and I use the tool to pressure test it, not to generate it from scratch. That's the version of AI use I think actually compounds over time.

But pressure is real. When deadlines get tight, I still sometimes fall into vibe coding, letting the AI just produce something and shipping it without fully sitting with the why. I'm not going to pretend that doesn't happen. The difference is that I notice it now, and I'm actively trying to make it the exception rather than the default. That's a slow habit to unlearn.

I make mistakes. I don't know a lot of things either. What I'm trying to hold onto is the constant trying, the constant noticing, the constant small correction. I don't think anyone has this fully figured out, and I'm not claiming to. I'm just trying to stay honest about where I actually am and keep moving from there.

So What About the Layoffs Talk?#

The framing that layoffs will be decided by who used more tokens misses what's actually at stake. If that were true, the safest people would be the ones who automated themselves out of needing to think at all, which is obviously backwards.

What actually seems to be happening, at least from what I've seen, is closer to what I wrote about in How Not to Be a Team in the AI Age. The people who become hard to replace aren't the ones avoiding AI or the ones using it the most. They're the ones who can still explain their reasoning when someone asks why. The ones who didn't outsource the part of the job that was never really about typing in the first place.

The conversation shouldn't be AI versus no AI. It should be: are you still doing the thinking, or did you quietly hand that off too?

That's the part nobody's FAFO-ing anymore. And I think that's the actual risk.

---

TL;DR: The “AI will replace coders” debate misses the point. Code was never the hard part, thinking always was. AI made it possible to skip the thinking, and a lot of people took that shortcut without noticing. The people who'll be fine aren't the ones who avoid AI or the ones who use it most, they're the ones who can still explain their reasoning. I'm not perfect at this either, but I'm trying to keep noticing when I slip and pull myself back.

---

If this resonated or you've felt something similar, I'd genuinely like to hear about it. Find me on GitHub, X, Peerlist, or LinkedIn.

Feedback welcome. Call out mistakes. I'd rather be corrected than stay wrong.