this post was submitted on 10 Nov 2026
264 points (97.1% liked)

Programming

29288 readers
393 users here now

Welcome to the main community in programming.dev! Feel free to post anything relating to programming here!

Cross posting is strongly encouraged in the instance. If you feel your post or another person's post makes sense in another community cross post into it.

Hope you enjoy the instance!

Rules

Rules

  • Follow the programming.dev instance rules
  • Keep content related to programming in some way
  • If you're posting long videos try to add in some form of tldr for those who don't want to watch videos

Wormhole

Follow the wormhole through a path of communities !webdev@programming.dev



founded 3 years ago
MODERATORS
 

I currently work at an "AI First" 🀒 fintech company, and we're mandated to program agentically. I was fortunate enough to start working before the AI hype, but the environment right now is daunting.

There's too much unmaintainable code slop and throw away scripts committed that I can't keep track anymore. Each PR has so much diffs that it feels like someone had diarrhea and shat all over.

I try to manually review the code, but it feels like best effort at most. Like I may only understand 70% of the thing that I'm reviewing. What ends up happening usually is that I just ask the model and verify the output(which is horrifying I know)

I spend the whole day arguing with AI, reviewing, and pushing back against its output that sometimes I wonder if I am the one driving AI or is it driving me.

--

Recently, I noticed that I've becoming more reliant on the model and that my skills have been atrophying. What would've taken me an afternoon if I program it by hand, would take me a week now if I do it now. I also find my head hurting and myself glossing over the code in code reviews and prompting claude to do even the basic of commands.

I get that there's more to programming in software engineering (collaboration, dealing with people, etc etc), but I don't feel like I am in control of the systems I am building anymore with the engineering aspect being replaced with babysitting and self lobotomy.

--

Off work hours, I try to brush up by reading on system design and architecture books, typing by hand if I'm learning a new technology or stack, but I feel like it's not enough to prevent from skill atrophy.

How are y'all coping and how do you recover from this? I am afraid on what skills that I'll be able to bring at my next place of employment if I resign and that I'm becoming a jack of all trades master of none type of thing - having only shallow understanding.

(page 2) 50 comments
sorted by: hot top controversial new old
[–] phoenixz@lemmy.ca 5 points 6 days ago

Easy!

Get fired from A CTO position because everyone gets replaced by AI, like me, and then be unable to get any IT job for 6 months straight until you're so desperate and that you'll take a call center job.

Oh wait, that isn't honing IT skills, that is honing depression skills. Sorry bout that.

[–] codeinabox@programming.dev 1 points 4 days ago

Which skill are you worried about atrophying? Writing code, reading code, or something else?

I find agents very useful for exploring, and understanding existing code bases I am not very familiar with, or if there are time constraints. One thing I have been thinking about, is using agents to assist with creating a plan, but instead of getting an agent to implement that plan, I'd rather make the code changes myself.

Another way could be regularly doing exercises at Exercism.

[–] WhiplashPost@programming.dev 5 points 6 days ago (1 children)

Personally I don't even work in anything that could be done with AI. All my tech stuff I do is hobbies. AI hasn't yet penetrated the in person manual labor sector much.

[–] boomzilla@programming.dev 0 points 5 days ago (2 children)

I switched from IT to manual labor. Working there for 3 month now. If I wouldn't have kept my fitness regiment while working in IT, I'd have no chance to stand my current job. Walking kilometers through 2 big halls overseeing and operating up to 5 machines using multiple of chemical agents in parallel and sometimes having to do manual labor that amounts to some solid workout sessions.

I still got my private Rust/Vue project and selfhosting stuff which won't see any slop till it's done. I do envy people working an office job a bit though. Even if they are bombarded with slop PRs. They don't know how good they have it if they haven't worked a job demanding full body commitment and getting up VERY early.

The work I do won't see any AI in the foreseeable future and they won't get a replacement for me without lots of investments as my predecessor estimated I'd need a year to learn all the stuff my job demands. I learned it in 2Β½ months. So a safe bank and good payment. But I yearn for a comfy office chair staring on code the whole day already.

load more comments (2 replies)
[–] ashughes@feddit.uk 2 points 5 days ago

By being unemployed

[–] ZILtoid1991@lemmy.world 2 points 5 days ago

Find an open-source project (better if it's noAI), then work on it manually.

[–] Taleya@aussie.zone 3 points 6 days ago

I throw it the mind numbing shit. Heres a bunch of sites, group them geographically for tech installs. Eat these logs and isolate element XX . Find instances of bullshit in this YAML. I keep the real work for myself.

Tried using claude for documentation, it was very clear the literacy and clarity rates of the files it was based on were dogshit. If you can't even write a process document for hbs sync that a drunken monkey can follow i got no use for you

[–] OnfireNFS@lemmy.world 3 points 6 days ago (1 children)

I setup a private GitLab instance for myself and have been doing some personal coding on the side. It's been refreshing. I even have used AI with it but I don't feel the pressure to ship and I have been reviewing and adjusting all the code it generates.

For work I've been getting closer and closer to "vibe coding". I added a testing framework and have the AI do test driven development and generate tests for the code it writes. That at least catches some of the bugs. But the expectation at work seems to have shifted to "if we ship bugs whatever. Well just issue a hotfix", "bugs used to make it into production before, why is it an issue if they do now".

When it comes to reviewing PRs they're 30,000 lines and it's exhausting to review those day in and day out. I've just stopped reviewing them, if something breaks it breaks.

My main concern with this is when something breaks that's actually important like security or payments related. In fintech that could be a very costly mistake.

I think having an ownership stake also makes a bit of a difference for me. Code I own and care about I actually take the time to refine. Code that if I leave I don't get to take with me is just code.

I like to think of it as designing mass produced furniture versus designing a custom woodworking project you are going to hand make. The mass produced one has deadlines, manufacturing will probably mess it up, it doesn't need to be perfect, at the end of the day you won't even get to own the design you came up with.

The woodworking project doesn't have a deadline, you can take as much time with it as you want, adjust it, re-cut the parts if they weren't perfect and you get to keep the blueprints you designed.

The woodworking project probably isn't scalable or financially viable but there's no doubt the end product is vastly superior

[–] kamstrup@programming.dev 4 points 6 days ago* (last edited 6 days ago) (1 children)

Mass produced furniture, like IKEA, goes through a very rigorous design process and have very high quality standards. They can spew out tonnes of product from an assembly line because the whole process is designed to be highly reproducible.

I think a slop factory is more like a restaurant that has decided to only cook using microwave ovens and decided that it's too costly to check if the meals are served raw or over cooked πŸ˜…

[–] OnfireNFS@lemmy.world 4 points 6 days ago

That's probably a better analogy. I actually didn't mention IKEA intentionally because their furniture is pretty good πŸ˜‚

[–] fonix232@fedia.io 3 points 6 days ago (1 children)

Enforce commit size and quality.

Label PRs:

  • xsmall for <25 LOC
  • small for 25-75
  • medium for 75-150
  • large for 150-250
  • xlarge for 250-500

Anything above and instantly reject. Do standard exclusions: docs, code comments, tooling (though I'd categorise this separately), etc.

Have very strict rules on creation of new scripts, new tooling, and so on.

[–] beejjorgensen@lemmy.sdf.org 3 points 6 days ago (1 children)

Instead of sizes, I recommend:

  • high quality for <25 LOC
  • quality for 25-75
  • decent for 75-150
  • questionable for 150-250
  • shit for 250-500
  • dogshit for 500+

Public histograms per engineer. πŸ˜„

[–] fonix232@fedia.io 2 points 6 days ago (1 children)

Not necessarily. Sometimes you keep features on separate branches and merge them at appropriate times, and those branches can grow BIG. But even if all individual PRs of the feature branch were reviewed, you still need to review the big blob.

Similarly, some things just take a lot of LoC without compromising quality.

e.g. in a project I work on, most API clients are generated (no, not genAI, but OpenAPI descriptors and Gradle tasks). When you add a new version, that can end up a big, chonky boi PR, but it's all auto mergeable because 1, it's generated via a known good template 2, CI already checked it.

[–] lucas@startrek.website 1 points 5 days ago

Gah, place I recently worked did this. So many conversations about why on earth generated code was being committed at all, polluting the repo and overcomplicating the diffs. Never got a good answer as to why they were committed, just "it's fine, skip over that in review". Then they went full on vibecode, too, which checked out. They just had no idea what they were doing.

[–] Zak@lemmy.world 3 points 6 days ago

I'm currently on a self-enforced break of at least a week from using LLMs for code generation. I'm using that time to pay off technical debt and make sure I have a full understanding of codebases that have grown too fast..

LLMs have let me get stuff done that's tedious or outside my skillset. I consider the former purely positive, but the latter is dangerous and requires being selective. I don't care to level up my Google Glass API expertise just so I can have heads-up biking navigation, but I do want to be good at maintaining the self-hosted compiler for the programming language I'm developing.

[–] danda@lemmy.zip 3 points 6 days ago

I've been working on my own templates and libraries for personal projects, and trying to do it all by hand to really learn the frameworks I'm using. It's been pretty rewarding and it feels like I'm tapping into those skills that I haven't used in a while. Focusing on templates/libraries also makes it not feel like throw-away work, as I will likely use it for something at some point.

[–] HelloRoot@lemy.lol 3 points 6 days ago* (last edited 5 days ago)

Programming slowly by hand with only a book (made of paper) for reference and no IDE.

EDIT: I mean in my free time as an excercise, not at work. Why the downvotes?

It was a big surprise to find out that I can't even write and compile a hello world from the top of my head without the IDE creating the demo project for me and the "play" button abstracting many steps away to run the app.

Oh and of course the decades of stackoverflow copy paste and delegating the grunt work to junior colleagues did nothing to help. I was athrophying long before AI.

I'm a CS instructor now (after two decades of industry), but it's unclear how long this job will exist despite my modernization efforts. When I think about what I'd do instead, one of the top things is working for a non-profit or public-sector gig doing Good and heavily leveraging AI to get the work done (using my SE skills to keep quality up). I figure it won't pay much, but I'm old enough now that the nature of the work has more value than the money.

And after the workday, I'll hopefully have something that I've never had in any other coding job: enough energy after work to code my own projects.

Hopefully.

If I had a manager who was telling me to review huge, unreviewable PRs every day, I'd just punt that to an AI with piles of tests. Under that management direction, there's nothing you can do to save the product; you can only postpone its inevitable demise.

[–] mx_smith@lemmy.world 1 points 5 days ago

Same situation as OP, when I do peer reviews using AI and it reports a list of flaws I just ignore it. If the Coderabbit doesn’t catch it then I’m just allowing the human error back into the code. It still sucks but there are future bugs to fix.

[–] Avicenna@programming.dev 1 points 5 days ago

use AI to learn new stuff making sure to check stuff that smells fishy.

Play Stationeers. It's got a neat little assembly language called IC10 inside that directly controls and networks the game's physical world and it's fun and challenging to master. Build contraptions, have deadly but hilarious accidents, repeat.

[–] Tollana1234567@lemmy.today 1 points 5 days ago

stop using AI?

[–] JeeBaiChow@lemmy.world 2 points 6 days ago

Side projects. I burnt out and left a client. Now rewriting an improved version of the tech that improves on every aspect of the system, much of which I developed in the fifteen or ao years before I joined.

[–] FishFace@piefed.social 1 points 6 days ago

I'm not really worried about skill atrophy. I don't really buy that using AI to write all my code for a month will do any long-lasting harm to my ability to write code more than not writing any code for a month would, and I've gone much much longer than that without writing code before, with no long-term consequences.

I think it's more about motivation, which again is something that you can regain pretty quickly if circumstances change.

[–] SouthernShao@lemmus.org 1 points 6 days ago

If you have strong soft skills this isn't so much of an issue, and I highly recommend that. AI can do such a good job in so many avenues, but being able to show up with confidence and understand and present materials to leadership is a very important skillset to possess. In addition, having the ability to solve problems on the fly is also a large asset from my perspective.

[–] epchris@programming.dev 1 points 6 days ago* (last edited 6 days ago)

I'm sorry that you're in this spot, it's not easy. I see a lot of what I love as a software engineer as "solving problems" and a big part of that was solving problems by writing code, but that part is changing.

But it sounds like you've identified a bunch of problems. I'm seeing how it feels to see these new frictions as a problem to be solved and finding fun in it so far.

Reviewing code wasn't ever real fun for me but it was managable at the pace we used to write code. So thats a new problem to solve: how can we confidently ship this new pace of code and feel confident about it? Some of that has been trying to design the process around it, some has been writing agents to help me stay sane while juggling the new amount of PRs I have to face.

It feels, so far, like trying to see all these new situations as problems to be solved, and experimenting with ways to solve them, has been fun for now, hopefully well find ways to keep it fulfilling.

[–] ultimaluz@lemmy.world -2 points 6 days ago

from the other side, i'm the cto of a small agency and i wrote our rules for working with ai, so maybe it helps. the first one is that the author answers for the code, "the model did it" is not an explanation for anyone. asking the model to review its own diff is a first pass, not a review, the same model that got it wrong is checking itself. the way i put it to the team, ai makes producing cheaper, not deciding or verifying, if you ship double with the same error rate you get double the errors. if your company only measures volume, the atrophy is more a management problem than yours, i'd push back with incidents and rework numbers, thats the language they listen to

[–] Wakko@hell.cloud 0 points 6 days ago* (last edited 6 days ago)

In short - I'm not.

This is a "buggy whip manufacturer" moment. Software development as we currently understand the practice is no longer needed in volume. Just like buggy whips no longer need as many craftsman making them since the advent of the internal combustion engine.

Yes, some people will still be "line editors" just like there are still some people preserving older skills and manufacturing niche items like buggy whips for the handful of horse-drawn buggies that still exist and are in-use.

But the vast majority of software going forward will be produced by LLMs. The skills required to "drive" the new LLM engine is an altogther different set of skills than what is required to whip and steer horses. Everyone fighting this tide is just fighting the inevitable.

Instead, I'm moving "up the stack" and focusing on answering the question, "What testing is difficult to game and shows me that the work has been done adequately to spec?"

The LLM-enabled future needs more mathematicians than it needs code-writers. The vast majority of anti-LLM Luddites I see online are all people who are "coders"; the "computer scientists" are busy adapting to a future that is already here.

load more comments
view more: β€Ή prev next β€Ί