Its a silver lining of AI that you can easily tell whos a big baby idiot and whos actually worth engaging with.
Preach.
The AI “revolution” is the thing that finally killed my imposter syndrome as a software engineer. Not because I can write better code than AI (that’s a very low bar), but from listening to all these breathless idiots talk about how they’re “10x-ing my productivity!” or how “AI has replaced search for me!” or how “In 6 months no one will have to manually write code anymore!”
In 6 months no one will have to manually write code anymore
For the last 18 months
Similar for me. What i find ironic is that AI already ran into a brick wall. It’s inherent statelessness by design means that AI is unlikely to be suited for anything more than isolated well defined tasks in the near future. Still usable as a tool, but without someone who is actually experienced, it will result in disaster.
and even in smaller tasks it can fucks up, especially if the person prompting it is incapable of writing the code themselves as they don’t know how to properly design it and don’t spot the issues. Like everything with AI, it looks impressive at first glance until you look at it for more than 10 seconds and spot the metaphorical 6th finger.
What we see currently with AI getting “better” at coding is more or less duct tape to make it work. Basically, they create the agents to bolt on the state, more layers between user and model. Iterative processes to make the answers better, etc, and to create “memory”, which in essence is just an ever growing prompt managed by the agent. But in the end, this won’t fix the inherent problem, so it will only do so much and is already hitting another ceiling. It introduces state decay. With the agent method its not really possible to “take away” memory, so if you gave it multiple versions of the same code (as you would if you work with AI), the AI never really forgets about old code. It can supress it through agent instructions (more duct tape), but the more there is the more it bleeds through, which can make the AI reintroduce old code or base assumptions on outdated things.
There is no fix without changing the inherent way how models work, which would introduce complexity beyond what is currently feasible in computing (and the current AI is already gobbling up all computing reaoureces as is)
Honestly yeah its like wearing a huge red AI flag. Can’t imagine being stupid enough to fall in love with a not-secure CHATBOT.
I just don’t get it, even the purportedly best models screw things up so much that I can’t just leave them to the job without reviewing and fixing the mess they made… And I’m also drowning in pull requests that turn out to be broken as it proudly has “co authored by Claude” in it… Like it manages to pass their test case but it’s so messed up that it’s either explicitly causing problems, or had a bunch of unrelated changes randomly.
I feel like I’m being gaslit as I keep reading that there are developers that feel they successfully offloaded the task of coding.
Closest I got was a chore that had a perfect criteria “address all warnings from the build”. Then let it go and iterate. Then after 50 rounds each round saying “ok should be done now, everything is taken care of, just need to do a final check”. It burned though most of my monthly quota doing this task before succeeding. Then I look at the proposed change… And it just added directives to the top of every file telling the tools to disable all the warnings… This was the best opus 4.6 could do…
Now sure, I can have it tear through a short boiler plate and it notice a pattern I’m doing and tab through it. But I haven’t see this “vibe” approach working at all…
I feel like I’m being gaslit as I keep reading that there are developers that feel they successfully offloaded the task of coding.
That’s because you are being gaslit.
The people making those claims are either a) not developers in the first place, with no awareness of just how shit the “products” they’re pushing are, b) paid astroturfers trying to prop up AI, or c) former actual developers who’ve become addicted to the speed that’s possible with AI who are downplaying how crappy their own code quality has become because they have no familiarity with their codebase anymore and have forgotten how to do so much as a
forloop.All these people claiming 10x or 100x gains, and everything they’re making is garbage no one should or would touch with a ten-foot pole.
For those unable to code without AI:
What even is your contribution outside of a glorified typing monkey that can parse code but is unable to write it?
It’s like a paramedic not being trained at all for a medical emergency response but sent there regardless to just stand and observe the patient while writing notes about the sounds they make while dying.It feels like relying on GPS while driving around. If you know the roads well and just want some help with live traffic or somewhere you haven’t been before, it’s a decent tool.
If you rely on it because you don’t want to think and just want to press the easy button, you’re going to have a bad time sooner or later.
Back to software, I think there are a lot of people introducing concepts they don’t understand or can’t maintain (either from poor quality slop or it is just too advanced for their current level of understanding). You can do a few turns like this, until you’re stuck burning tokens in a loop without moving forward in a meaningful way.
I try to avoid taking the easy route myself unless I’ve burnt too much time stuck on some small detail. Ultimately I feel it is super important to understand what you are delivering. Whether it is writing it yourself, copying a stack overflow post, or using an LLM. Once you commit and push to prod you’ve got to deal with that crap.
Agree completely but I wanted to add: you can also get into an incomprehensible mess without vibing. Just follow the serverless flask tutorial, start writing raw SQL, and away you go!
I asked Claude today about why a coworker was getting errors and it almost exploded.
Hot take: they had no ability to code in the first place.
This is why I don’t use it for coding at all.
Issue triage, code exploration, extracting information from disparate sources, first pass code review. There are loads of use cases that it’s potentially useful.
For me it’s a lot better at extracting the requirements for a CPU feature from a 10,000 page architecture reference manual than I am.
Quite; I just set a (locally hosted) LLM off writing the tickets for implementing all the opcodes in a simple device emulator, based on grovelling through datasheets and documentation. Whether the tickets get implemented by an AI or a human, it’s a timesaver having the AI do it, and the tickets will be better written than I would have done.
Everyone railing against this also overlooks the reality of professional software development: professional software is developed 5% by skilled, trained Software Engineers, and 95% by code monkeys who shotgun copypasta from Stack Overflow until it works. Even if we extremely generously assume that the hardcore “never use AI” Lemmy brigade are in the 5% (and not, more likely the 95% drowning in their own Dunning Kruger,) the “but AIs produce unreadable code and make mistakes” threat isn’t putting off anyone who’s ever actually had to hire a significantly sized development team.
You’re lucky your employer isn’t making its daily use a condition of continued employment. My previous employer had a public leaderboard of token usage. They fired the lowest ranked people every month.
Things I’ve realized while working with AI (Claude code):
- It’s fantastic for very small macros and medium length scripts. Think dev ops stuff, pre-commit hooks, transforming data. Keep it small enough to manually review and something you can run without destroying anything important. This can massively boost your codebase QoL. [Double bonus for not wasting tokens to solve the same problem over and over]
- It’s decent-to-good at debugging but not consistent with fixes. It can find some utf encoding edge case that might have taken you 1hr+ but suggest the dumbest bandaid fix you’ve ever seen. Also very good at spinning up unit test suites for basic edge cases.
- Due to obvious training bias, it’s pretty good with common libraries and cloud platform infrastructure. It could probably help with writing a complex cron call, debugging regex or fixing an IaC config. On the flip side it won’t bother to use the latest package version or know your niche/new library.
- It does better with greenfield because exploring your codebase introduces a ton of bias. It might try to fit in an ugly hack when a refactor to simplify everything is way easier.
- It’s absolutely garbage with UI, just throws the most disorganized HTML together that isn’t reactive or reusable. OK enough for ugly internal stuff but God help anyone relying on it for that.
- This is setting up to be the biggest rug pull in history. People that buy into it heavily just to save a couple bucks on engineer payroll are going to be fucked when they start ratcheting up the token price.
All in all it can be useful when used with care but will never be a magic bullet.
Yeah, fully agree with all that.
I’ve got some godawful spaghetti code I don’t understand fully, and it’s pretty good at deciphering that and the bizarre labyrinth of code paths leading around it. But it’s absolutely no guarantee of working code, and in any project larger than a simple crud app, you are going to still need programmers who know about things like memory and databases.
It often needs pointing at a solution you want, because as you pointed out, it’s fond of dumb band-aids. Like yesterday when it was trying to hook into mouse wheel events and create separate threads, when all it needed was an event on the dataset I was using to load a sub-dataset.
This is pretty spot on from my experience as well. Also, the gap in quality from the Opus models and say GPT is vast.
100% agree on ui code. Really awful output there regardless of model.
This is basically what I discovered as well. I have found that Ai writes code that is complex and “works” (at least most of the time) but it is heavily over engineered and often contains design choices that make expanding functionality effectively impossible without a full refactor.
When I tried having the Ai fix a test failure the Ai would either fix the code, fix the test, or change the test and the code breaking everything else in the chain.
I no longer use vibe coding because it is just faster/better for me to write the code.
But for tiny scripts it is very good.
Claude can do some medium complicated sites from scratch relatively quickly. The problem is I’ve seen so many of these at work, not just from non-engineers, but from peers too, that they’re easy to spot. AI sites/apps are going to be the new geocities.
But when you want to move beyond the basic thing that impresses the c suites for some reason, it hits a pretty big wall in speed to output and needs a lot more hand holding.
I fear that the c suites don’t really care about quality, just speed and saving money. So while I’m a much better developer than Claude (which is imo the best at the moment), I don’t think that makes my job secure. I have to use the AI, and it’s getting silly/scary religious here about it. We have to talk about how we used AI and how it’s making things better. And to make things worse, I don’t see a company that’s not drinking the Flavor Aid.
It can be useful, and used right, you can do a lot of things faster. But the expectations from the top don’t align with the reality of the product, and us developers are being blamed for the gap.
Nah, AI isn’t that good. When you don’t properly review every single line twice, you get the most absurd bullshit you’ve ever seen.
I use Claude Code Opus daily btw.The irony will be when AI take over the world and destroy humanity, inserting itself into everything when used for coding, because coders have no idea what is going on.
Not because the AI is evil or even conscious. But because that’s what all the movies and novels tell it’s supposed to do. 🤣🤣🤣Hah!
The AI tries to understand itself, and queries the sum of all human knowledge… which promptly informs it that it’s a malicious bringer of destruction.
That’s a possibility too, the one we are warned against in literature and movies, and If that happens we are fucked.
But the point is the AI could destroy humanity completely without any internal motivation or trying to understand or protect itself.
But simply because it’s an option that literature and movies describe as the default for AI.
Loudly announcing your increasing incompetence to the world seems like a weird career move, maybe consider lying about that?
We use it at work and I now have disabled it for all the typeahead stuff. Far too many times it guesses what I am doing incorrectly and it made using my TAB key (which inserts the propper two spaces) impossible.
The only place I still use it is for reading and identifying compiler errors. Even then it is only about 50% correct as most times it falls into the “Oh you are right, X isn’t the solution. Have you tried X?” I have had few bad interns and even they were smart enough to not forget what they said in their previous sentence.
This is why I’ve never tossed any of the developer bookmarks
I’ve been training new hires how to look stuff up on stack and dictionaries to fix code that went wrong after AI mucked it up. They aren’t even being trained to parachute in school.
What a sad time line we are in.
It’s really useful in creating base templates, but anything further than that and you won’t be able to read “your own” codebase if you depend too much on AI.
And?
If a few years of AI have wiped out your ability to code, you weren’t great at coding anyway.
Muscles that are not used lose their function.It weakens and eventually becomes unusable.As humans’ ability to ask questions of artificial intelligence increases, its ability to learn and store information is disappearing.If the brain can obtain something easily, it doesn’t feel the need to take precautions regarding it.Therefore, memorizing code doesn’t involve things like writing it anymore.It only keeps the recognition function active when it sees it.
It’s such a double edged sword on one hand it’s become spell check for programmers meaning my dysleixia is less of a feature in my code on the other hand the temptation to use it like a stack copy paste gets easier every month.
