Do programmers even feel like programmers anymore?
by•
Saw a post from an 8-year dev this week. He said any junior with an agent now matches his output, and he feels no pride left in it. It got huge, because a lot of people quietly feel the same.
I read it from the other side. I'm a designer, not an engineer. For years "can you build this" meant "find someone technical and pray they stay interested." Usually they didn't. AI took the thing that hurt that dev and handed it to me.
Same shift. Loss from one seat, freedom from another.
I'm shipping a Mac app solo right now, as someone who can't write a backend from scratch. Two years ago that was a joke.
So I'm curious about the makers here. If you build for a living, does it still feel like your craft, or like something else now? Loss or relief?
397 views
Replies
I think it still feels like a craft, but the craft has moved up a layer.
Before, a lot of pride came from writing every line yourself. Now the real skill is knowing what to build, how to structure it, how to review the output, and how to avoid creating a messy system that only works in the demo.
AI makes building more accessible, which is a good thing. But taste, judgment, debugging, architecture, and responsibility still matter a lot.
SlimSnap
@prashant_patil14 "Moved up a layer" is right. The uncomfortable part is the new layer is less teachable. You can grind syntax. Taste and judgment you mostly earn by shipping and being wrong. AI made that skill scarce, not cheap.
SlimSnap
@kevinpickles Exposure being the only teacher is the scary part for juniors. The grunt work that used to build that judgment, the boring tickets, the debugging, is exactly what AI now does. The ladder to judgment lost its first few rungs.
the craft didn't move. the tools did. i used to feel like a founder because i could write the code. now i feel like a founder because i decide what to build. both are just versions of "you get to choose." label changed. ownership didn't.
honest answer to loss or relief: relief. the parts i loved (shaping, editing, deciding) got easier. the parts i tolerated (setup, boilerplate, deploys) got faster. i don't miss the friction. i miss thinking i was the only one who could beat it.
SlimSnap
@thenameisarian "I miss thinking I was the only one who could beat it" is the whole thread in one line. That's not losing the craft. It's losing the moat around it. The work you loved is still yours, the exclusivity isn't. Those feel the same at first, and they really aren't.
I have been developing and writing small projects for the last 10 years. Nothing major, just enough to keep learning, experimenting and building things that interested me. Over the last year, AI-assisted development has changed the way I work. For me, the principles have not changed, but the tools definitely have.
In the past, my usual path was Reddit, Stack Overflow, forums, documentation and a fair amount of trial and error. Now, a lot of that starts with a chat prompt across different AI tools and agents.
But the fundamentals still matter.
Secure development, testing, good judgement, curiosity and common sense have not gone away. If anything, they matter even more. AI can help accelerate the process, explain concepts, unblock problems and give you more confidence, but it should not replace experience, careful thinking or the passion for building something properly.
SlimSnap
@francois_marais_nz The Reddit and Stack Overflow to chat-prompt shift is the part nobody mentions. That whole "search, read five threads, stitch an answer" muscle just went quiet. And you're right that it raises the bar on judgment, not lowers it. When the code is basically free, telling good output from a demo-only mess becomes the actual job.
Curious when you say you're shipping a Mac app solo what parts is AI actually doing for you ?Coding? Architecture? Debugging? Or mostly helping you move faster?
SlimSnap
@alan_gregory Most of the coding. I studied CS, so I can read it and tell when the architecture's off, but I work as a designer and never shipped a real product solo before this. The agent writes the bulk, I make the product and UX calls and push back when something feels wrong. Debugging is where it saves me most. That used to stop me dead.
The craft hasn't disappeared it has shifted. We're spending less time typing code and more time thinking about architecture, workflows, and user experience.
As a Designer I ponder the same question everyday.
I more vibe prototyping than designing. Should I vibe-code and learn or learn and vibecode.
Do I need to know everything about the tech? I guess not.
I feel soon, the term designer will be droped and give way to something like Experience Builder or Product Expereince .
SlimSnap
@roopesh_donde You don't need to know everything about the tech. You need to know enough to catch when the output is wrong. That's a small thing to learn and you get it by building, not studying first. The title question I care about way less.
@bickov Yep, thats the approach I am following right now. Learning by building.
The part nobody's mentioned yet: this also broke hiring. A take-home test used to be a decent signal on its own. Now it tells you almost nothing, because the output looks the same whether someone reasoned through it or just accepted whatever the agent handed back. We redesigned interviews around live pairing where we watch someone review and push back on AI-generated code, not write it from scratch. It's a genuinely different skill to evaluate, and most of our old rubric didn't map to it at all.
SlimSnap
@galdayan Live pairing to watch someone push back on AI code is smart. Writing from scratch tests a thing that barely happens on the job now. The hard part is scoring it though. How do you tell real judgment from someone who just sounds confident saying no?
As in any profession, we are given tools and use them. What's disturbing is the ratio of AI to human work. After almost 25 years of programming, I often find myself solving simple things with AI prompts, while simultaneously forgetting how things were done in the pre-AI era.
SlimSnap
@ukash Forgetting how you did it pre-AI is only a problem if you also lose the judgment to catch a bad answer. The syntax you can relearn in a day. The instinct for when the output is wrong is the part worth protecting. People mix up those two.
@bickov True! I'm working on codebase that have > 12m lines (excluding vendor packages), it's a lot. There isn't any AI that is able to understand full context, architecture or relations between components in project of this size. Because of that, old-school human made PR review are still what we need. For now human-dev way is safe :)
SlimSnap
@ukash The size thing is the real ceiling right now. On a small greenfield app the agent holds the whole picture. At 12m lines it can't, so it guesses the relations and you catch that in review. But "for now" is carrying a lot of weight in your last line. That context limit has moved every few months. I wouldn't pin the safety on it staying put.
Relief more than loss. Spending hours chasing syntax errors never felt like craftsmanship anyway. Time is better spent thinking through edge cases and making better decisions.
SlimSnap
@jason_scott8 Syntax was never the fun part. The decisions were. Losing the syntax grind doesn't read as losing craft to me, more like someone finally deleted the boring layer.
For me, it still feels like programming—just a different version of it. AI writes some code, but deciding what to build and making everything work together is till the real challenge.