I’ve found that the quality of code produced by my team has dropped significantly since my org pushed everyone to use LLMs. “Let me throw an agent at it” usually means “here’s another headache of a PR for me to review.” Engineers with decades of experience are writing such utter fucking crap now, and it doesn’t matter how many docs I write for their fuckass agents to ingest, the quality doesn’t seem to be getting better. They can make their agent loop on it for functionality/bug and hallucination squashing, but that usually just means I get a bloated fucking mess to review. It’s miserable and it makes me want to go shovel actual hog shit.
Sincerely,
a very sad and tired senior dev who used to enjoy his job
I’m trying to adjust expectations. You aren’t getting my approval unless you can explain why each part is there. Your poor planning is not my emergency. K.I.S.S. still applies.
Admittedly, this is easier to say as a senior whose opinion is respected within the company.
As a senior dev… AI actually improved the code we wrote in my last job. Mind you the codebase was in a shite position (tons of technical debt, plague-like avoidance of inline comments because “all of that knowledge is on Jira/Notion” except it wasn’t; lots of legacy code, undocumented weird behaviours, everyone was doing their own coding style without regards to the rest of the codebase and so on).
With controlled AI contributions, we’ve managed to:
refactor large chunks of code where the overall high level logic didn’t need to change, but the underlying tooling did (e.g. lots of UI layers got rewritten from a very early attempt at a standardised MVVM approach to a well defined MVI one, we’ve also managed to eliminate the remaining RxJava/RxKotlin legacy flows to replace them with Coroutines)
UI layers got refactored in no time because the AI agent was able to convert them appropriately (and even fixed a handful of UI bugs)
a handful of core monoliths (yes I know… I hated working with them as well, but we never had the chance to properly refactor) got split up into more manageable, vertical-specific domain controllers without breaking
the elimination of a legacy overarching component that was estimated at 6 months, got reduced to 6 weeks
unit test coverage skyrocketed because even if the code was hand-written, an agent would provide coverage, based not just on implementation but on intention (and adoption of TDD/BDD finally started spreading).
Meanwhile code quality improved because we’ve had well established patterns and styles defined in the agent instructions, and even the anti-documentation “pragmatists” got shut up when they finally understood that the AI works better if there’s fucking comments to explain what’s going on
What you’ve done is allowed an unethical and incompetent machine create an unmaintainable technical debt which will destroy the future of the codebase or the company, because you were all lazy.
He let AI refactor his entire codebase because the team didn’t understand each other’s techniques/approach and nobody had any time or knowledge to do the work themselves. They literally did not understand how it worked so they asked the chatbot to redo it and now they still don’t understand it but it appears to them as if it was improved.
IDGAF if a slopper is checking the slop. Since 2022 the number of bugs and exploits has exploded dramatically and code is measurably worse: because people do not understand code that they did not write.
You do realise that most of those bugs and exploits aren’t from AI generated code (a good chunk has existed for years before LLMs were even a wet thought), and the number is mainly exploding because LLMs are being used to identify them, right?
Take the highly coveted, human audited OS OpenBSD. 30+ years of code, deemed to be the most secure, most closely watched and verified on the planet… And a simple LLM check identifies not one, not two, but three (by my last count, this could’ve increased since, I don’t follow OpenBSD that closely) major, exploitable issues that humans have missed for years.
You do realise that most of those bugs and exploits aren’t from AI generated code (a good chunk has existed for years before LLMs were even a wet thought), and the number is mainly exploding because LLMs are being used to identify them, right?
I think that’s the lie you’re being fed, mate. Sure, some of these old existing exploits have been identified by these multibillion dollar tech companies, but we’ve well documented that AI slop coders produce terribly unsecure code and far more than their peers.
Hey now, not all of us are spouting nonsense constantly. I for one tend to research topics - even if in a cursory manner (and no that does not mean I use Cursor or other AI as my main information source, before another anti-AI shitwit accuses me of doing so). And I tend to stick to topics I have at least a cursory understanding of, to begin with.
Should that not be the case, I’ll try to join the discussion with questions. Problem is, what I consider the basics of a discussion worth having - that is, being informative, with the goal of education, sharing knowledge - seems to be too much effort for most.
However what I can’t stand is the aforementioned shitwits joining a level headed discussion only to barf in a half-digested word salad based on nothing but their emotions. If I wanted to debate based on emotions, I’d go to a reality show community…
Civility goes out the door the moment you decide to not actually contribute anything useful but regurgitate an “argument” of pure emotion without any factual basis.
So I’ll repeat, fuck the fuck off, and when you think you’re far enough, fuck off a little bit more.
I’ve found that the quality of code produced by my team has dropped significantly since my org pushed everyone to use LLMs. “Let me throw an agent at it” usually means “here’s another headache of a PR for me to review.” Engineers with decades of experience are writing such utter fucking crap now, and it doesn’t matter how many docs I write for their fuckass agents to ingest, the quality doesn’t seem to be getting better. They can make their agent loop on it for functionality/bug and hallucination squashing, but that usually just means I get a bloated fucking mess to review. It’s miserable and it makes me want to go shovel actual hog shit.
Sincerely,
a very sad and tired senior dev who used to enjoy his job
If they push shit, I push back. It’s a tool they can use, not an excuse. Creating rules for the agent is fine, but that’s only a soft guardrail.
We don’t have a mandate to use LLMs, if things are worse with them then we won’t use them.
That’s the mindset we started with but things have shifted to what we’re seeing now and people are being forced to use it in many cases.
I’m trying to adjust expectations. You aren’t getting my approval unless you can explain why each part is there. Your poor planning is not my emergency. K.I.S.S. still applies.
Admittedly, this is easier to say as a senior whose opinion is respected within the company.
As a senior dev… AI actually improved the code we wrote in my last job. Mind you the codebase was in a shite position (tons of technical debt, plague-like avoidance of inline comments because “all of that knowledge is on Jira/Notion” except it wasn’t; lots of legacy code, undocumented weird behaviours, everyone was doing their own coding style without regards to the rest of the codebase and so on).
With controlled AI contributions, we’ve managed to:
Meanwhile code quality improved because we’ve had well established patterns and styles defined in the agent instructions, and even the anti-documentation “pragmatists” got shut up when they finally understood that the AI works better if there’s fucking comments to explain what’s going on
What you’ve done is allowed an unethical and incompetent machine create an unmaintainable technical debt which will destroy the future of the codebase or the company, because you were all lazy.
It sounds more like you didn’t even read his comment and are just rabidly spewing anti-AI talking points instead of participating in the discussion.
He let AI refactor his entire codebase because the team didn’t understand each other’s techniques/approach and nobody had any time or knowledge to do the work themselves. They literally did not understand how it worked so they asked the chatbot to redo it and now they still don’t understand it but it appears to them as if it was improved.
What did you read?
Literally every single statement you made includes an assumption that the work wasn’t checked and that nobody has any idea what it changed.
Yes, slop exists, but one-shotting without review is not how most people use the tools.
IDGAF if a slopper is checking the slop. Since 2022 the number of bugs and exploits has exploded dramatically and code is measurably worse: because people do not understand code that they did not write.
You do realise that most of those bugs and exploits aren’t from AI generated code (a good chunk has existed for years before LLMs were even a wet thought), and the number is mainly exploding because LLMs are being used to identify them, right?
Take the highly coveted, human audited OS OpenBSD. 30+ years of code, deemed to be the most secure, most closely watched and verified on the planet… And a simple LLM check identifies not one, not two, but three (by my last count, this could’ve increased since, I don’t follow OpenBSD that closely) major, exploitable issues that humans have missed for years.
But sure, slop.
I think that’s the lie you’re being fed, mate. Sure, some of these old existing exploits have been identified by these multibillion dollar tech companies, but we’ve well documented that AI slop coders produce terribly unsecure code and far more than their peers.
Hilarious 😂🤣😂🤣 just listen to yourself
As they usually do.
Chill out dude
Please shut the fuck up if you have no inkling about what you’re talking about.
On the one hand that would be nice, but on the other hand it would result in basically no content on social media any more.
Actually, no, I’m on board with this.
Hey now, not all of us are spouting nonsense constantly. I for one tend to research topics - even if in a cursory manner (and no that does not mean I use Cursor or other AI as my main information source, before another anti-AI shitwit accuses me of doing so). And I tend to stick to topics I have at least a cursory understanding of, to begin with.
Should that not be the case, I’ll try to join the discussion with questions. Problem is, what I consider the basics of a discussion worth having - that is, being informative, with the goal of education, sharing knowledge - seems to be too much effort for most.
However what I can’t stand is the aforementioned shitwits joining a level headed discussion only to barf in a half-digested word salad based on nothing but their emotions. If I wanted to debate based on emotions, I’d go to a reality show community…
Eat my ass slopper
Yeeeeeah, not gonna listen to some fuckwit who immediately launches into an unprompted tirade without even reading what they reply to.
Oh I’m sorry did I not enter this discussion with enough civility?
Civility goes out the door the moment you decide to not actually contribute anything useful but regurgitate an “argument” of pure emotion without any factual basis.
So I’ll repeat, fuck the fuck off, and when you think you’re far enough, fuck off a little bit more.
WHOMP WHOMP