The sad reality is that most apps are regurgitated CRUD with new design language. The llms are really good are regurgitating this stuff, so the only interesting part left is the 10% that was already interesting before. I just take 2h lunches and spend time with my family while the llms puke out the CRUD, then thoughtfully work through the 10%
AI code assistants currently are like that senior dev who knows everything, has experience in everything, and always likes to help everyone. But sadly he does a lot of shrooms in his spare time and sees things that aren’t there. He also has quite the opinionated ideas about what good code should look like.
That implies that you just can’t let him do the quality assurance. He can write code. You can ask him stuff. And he excels at debugging.
But you can’t let him plan the architecture. You can ask him about his ideas and select the ones that are actually good, though.
And you definitely can’t let him review his own code. Someone else has to do that.So if they literally just let the AI do everything, there is absolutely no way anyone has a chance to ever review the result because AI code is harder to review (for now), and AI loves to repeat itself (for now).
And sure, they can tell it to write tests. But AI is dogshit at writing tests. It tends to forget the important cases, test the wrong thing, and/or obsess about completely irrelevant implementation details.
I also tried the multi-round approach with a committee of agents trying to iteratively find and fix flaws. It only works well for finding bugs, not fixing them (for now).But while AI can’t replace me, it is really nice to have as a tool - just like I have syntax highlighting in the IDE, I have a code assistant that can autocomplete a few lines of code, find out why my code misbehaves, or tell me what some library method is supposed to do. Sure, I have to still verify the results. But often, finding the bug or information was the hard part.
Coding assistants are great. Just don’t let them write too much code at once. You need someone who still actually understands the codebase. That sadly can’t be an AI (yet).AI code assistants currently are like that senior dev
More like a junior dev.
I really love when I ask the AI to do something, and it gives me 40-50 lines to do it, and I tell it there’s no way something like that should take that much code, and it’s like oh, ya you’re right, here’s a 5 line version using something from from the standard sdk.
I’ve tried to put in its memory to not do things like that, but it keeps doing it.
I guess the AI isn’t iterating over its “ideas” and code yet. I come up with overcomplicated and overcomplex code too. But my natural neural network keeps reevaluating the code while I think about it, write it down, test it, and often even come back to it later after I moved on to another part of the codebase.
It looks to me like AI is using the infamous waterfall method to write code. It does the “thinking” phase and does bugfix iterations when compilation or tests fail. But never does it question the design it came up with after it wrote it down. It doesn’t reevaluate whether the last round of fixes made tests work that test irrelevant implementation details. It doesn’t question whether there really need to be four methods sharing 90% of their body for doing the basically the same thing only subtly different.
I hold the suspicion that AI will be as slow as humans when it becomes as reliable because what makes us reliable is the redundancy in our thinking processes. The constant reevaluation in subtly different states of mind helps to catch design flaws and silly mistakes.Reevaluation of finalized code is probably not something you can just put into memory. You would need to modify the harness to automatically inject a new prompt triggering reevaluation of the design and implementation after each step. This would likely increase token use and execution time by an order of magnitude. And if you do it by spawning fresh agents, there would be a high chance of having the AI go in circles, switching between a few different versions of the same shitty design or implementation. So you need to keep the context, which grows each iteration. So context rot may become a real problem. The AI might deviate from the initial goal just like some human devs tend to do…
And natural neural networks learn on the job. They adapt to the codebase and industry they are working with/in. claude.md and memory text files help a lot. But they are still sometimes ignored. And I don’t know why. Also, human rules are almost always somewhat fuzzy. Long lines are to be avoided - but unit tests and that one massive dict of dicts really are a lot more readable when ignoring that rule. There might be a lack of good training material on what makes code readable for humans.
That actually makes a lot of sense and goes with how ill code something larger as well.
Get a rough version working, sometimes horrible code. See how it functions. Reassess it and optimize it. Go bug hunting, which will often result in more optimizing, then finish all the error handling I probably skipped and only logged. More bug hunting, then review my code like a code review before submitting it for a code review.
Possibly motivated to do more complex problems to spend more tokens?
I don’t hate Claude Code or whatever. I hate the company pushing you to constantly churn out shit so you need to turn to these tools to keep up pace or get fired.
Architecture decisions or code quality? Fuck it just ship.
You have concerns about an approach? You’re the enemy to your coworker’s promotion. Get out of the way.
It’s all just so depressing. It will all come crumbling down. It’s a matter of when not if. But god I hate this ride right now.
Time to lean into the slop extra hard. Use AI to create massive PRs, use AI to approve the PRs. Automate sending slack alerts when a feature ships. Be loud, really show some massive “Productivity”.
Our place measures PR counts so a ton of tiny ass PRs is actually the move.
I’m sorry for your shitty workplace.
Yeah it’s not fun. Between that and now quarterly performance reviews things have really gone downhill.
Quarterly? Wow. Some MBA was probably like public companies need to report quarterly so we should put employees on the same schedule! Dumbass idea.
It’s so bad. It’s an absolute waste of time and resources.
Leadership is like “just use AI to speed things up”. But if I’m just pasting slop in and my manager is using slop to parse and give feedback what’s the goddamn point???
My slop machine says you’re not slopping slopply enough! Get back to slopping, and we’ll slopback in few months!
People are working 12 to 13 hours a day
yeah I’d write a diatribe about the best work in the world under those circumstances too
The coder does the first 4 hours, Claude does the next 9. You team up the next day to review each others work at lunch. Log 13 hours on your timesheets. 14, because you worked during lunch.
workers spend 12-hour days pressing enter
If only that could be automated somehow.
I’ve heard the same from developers at work “doing this using AI makes it boring, but at the same time I feel silly if I write everything from scratch”.
If I’m forced into Claude Code or something similar, I will be looking for a new job. Ideally a different role in my company to keep my benefits and not have to completely start over, but if I have to jump ship, I will. Fortunately, our security and GRC teams have explicitly blocked Claude Code for the time being. I still have the tools built into Visual Studio, but I seldomly use them, and when I do it’s basically as stack overflow, and that only when I can’t find what I’m looking for through web searches first.
We had someone from our Networking team come over the other day and say that he could tell our summer interns were gone because our AI usage dropped by over $1000 / month.
Look I’m all for the AI hate but this is just stupid. I’m aware this might not be the most popular opinion here but be warned
Honestly if I were to change jobs atm asking about their AI policy and usage would be one of the first questions that I’d do and if they don’t use it well and haven’t figured out how to integrate it in their company (as a tool, not necessarily as a product) then I’d end the meeting then and there.
I’d quit my current company if they prohibited AI tools. Fortunately they don’t force us to use them, nor but we’re all encouraged to find actually uses for it (which in Software Engineering are a ton). If one of my colleagues wouldn’t use AI in any way I no only would consider them less productive, I’d consider them a worse engineer because it shows they can’t adapt.
AI in general and Claude Code as an example is an AMAZING tool. You just gotta use it properly. I still own the code I ship even if I understand it in less detail than before (that’s the trade off by being able to ship many more features than I could in the past).
I’ve been in this company for a few years now and just in the past months there are so many projects I have done that there was never an opportunity to do so ever since I joined because it would be days/weeks of work and now it’s a couple of hours while I work on other shit.
And people/companies that spend hundreds of thousands of dollars in AI every month are idiots who are just throwing shit at a wall and seeing what sticks. My company pays me a 100$ month subscription and many colleagues have even cheaper ones. Thats like 1% of my salary and it for sure increases my productivity many orders of magnitude of that.
Nah, fuck AI. I will pivot careers. Anyone that claims to be a developer but generates all of their code is just a modern day script kiddie. Idgaf about efficiencies because the ethical implications of AI are far worse than the extra money I can make for a company.
I will never be an AI code reviewer, because that’s all prompting is. Like I said, I’ll pivot careers out of dev work before I do that. Pick up a trade and work until I’m 80 if I have to, but that’s still better than being one of Claude’s bitches.
Outside of the ethics of AI, the idea of sitting in front of a screen and being a “prompt engineer” takes everything I enjoy out of coding. If it was that or death, I’d choose death.
Replace all of that but instead of AI use computers or calculators.
I’m sorry but that is absolutely deranged talk. And you won’t have a choice about pivoting careers. You will be forced to in the same way that nobody in this millennium will hire someone that does their accounting with an abacus. And that’s a good thing.
Now, am I happy that a lot of people will be out of a job and the profits of that will be funneled into a few pockets? Absolutely not, but this doomsday, Amish like talk won’t help that conversation because the problem isn’t the technology. It’s who owns it
I have not found AI code to be so helpful. I actually use it a lot less and more strategically than back when it first became viable. It can save a lot of time in the right situation though.
Honestly if you don’t find AI code to be helpful that tells me more about you than AI.
It is incredibly helpful barring some very specific niches
Oh it’s way worser than that. Its erasing our skill
If you’re just hitting enter you’re doing it so wrong.
Agents are brain gang banging him.
They just need that fancy controller openAI made. Then it’ll be an approve button instead.
oh no a tech worker is whining about something again. Quick everyone rush to their defense. They must remain perfectly comfortable and happy inside their air conditioned luxury office.
Its work and you’re paid well for it. Suck it up or put your skills to use at a different company.
Your tongue is so good at cleaning boots. Very well done!
The guy is working at a shitty sweat shop and blame AI for his misery instead of moving his ass to another job.
Let’s ask Claude to write an article about it.
One… use --dangerously… they asked for speed, thats how you get it. Two… if you have great tests, you don’t really need to know every detail of the code. The risk just shifts from a dev designing bad code, to a dev designing bad tests.
I dunno I’ve seen too many times where Claude Code will just rewrite a test it doesn’t like and say it’s good. Combining that with —dangerously or —yolo is a recipe for disaster.
