…and I still don’t get it. I paid for a month of Pro to try it out, and it is consistently and confidently producing subtly broken junk. I had tried doing this before in the past, but gave up because it didn’t work well. I thought that maybe this time it would be far along enough to be useful.
The task was relatively simple, and it involved doing some 3d math. The solutions it generated were almost write every time, but critically broken in subtle ways, and any attempt to fix the problems would either introduce new bugs, or regress with old bugs.
I spent nearly the whole day yesterday going back and forth with it, and felt like I was in a mental fog. It wasn’t until I had a full night’s sleep and reviewed the chat log this morning until I realized how much I was going in circles. I tried prompting a bit more today, but stopped when it kept doing the same crap.
The worst part of this is that, through out all of this, Claude was confidently responding. When I said there was a bug, it would “fix” the bug, and provide a confident explanation of what was wrong… Except it was clearly bullshit because it didn’t work.
I still want to keep an open mind. Is anyone having success with these tools? Is there a special way to prompt it? Would I get better results during certain hours of the day?
For reference, I used Opus 4.6 Extended.
producing subtly broken junk
The difference between you and people that say it’s amazing is that you are capable of discerning this reality.
What I don’t get, though, is how the vibe code bros can’t discern this reality.
How can they sit there and not see that their vibe-coded app just doesn’t do what they wanted it to do? Eventually, you’ve got to try actually running the app, right? And how do you keep drinking the AI kool-aid when you find out that the app doesn’t work?
They’re the same people that copied code from stack overflow that you had to tell them how to actually fix every PR. The difference is the C suite types are backing them this time
Vibe code bros aren’t real programmers. They’re business people, not computer people. Even if they have a CS degree, they only got that because they think it’ll get them more money. They lack passion and they don’t care about understanding anything. They probably don’t even care about what they’re generating beyond its potential to be used in a grift.
I graduated college not that long ago and my CS classes had quite a few former business majors. They switched because they think it’ll be more lucrative for them but since they only care about money they didn’t bother to actually learn the material especially since they could just vibe code through everything.
So much this.
After working in tech companies for the last 10 years I’ve noticed the difference between people that “generate code” and those that engineer code.
My worry about the industry is that vibe coding gives the code generators the ability to generate even more code. The engineers (even those that use vibe tools) are not engineering as much code by volume compared to “the generators”.
My hope is that this is one of those “short term gain, long term pain” things that might self correct in a couple of years 🤞.
It’s insane that companies are going back to metrics like LOC (or tokens generated), when the industry figured out decades ago that these are horrible, counterproductive metrics.
“The hard thing about building software is deciding what one wants to say, not saying it. No facilitation of expression can give more than marginal gains.” - No Silver Bullet (1986)
I do apps that work, i do patches that are production quality. Half the cs world does… I do full stack ai debugging of esp32 projects.
It’s a powerful tool, you just need to learn it’s strong and weak points, just like any other tool you use.
Half the cs world does…
What’s the basis for this claim? I’m doubtful, but don’t have wide data for this.
Rough estimate from my personal connections only. Some work places where ai is not possible, but all that have made an effort report good code. You need to work with what it is - a word generator that sometimes gives correct results. Make it research and not trust training. Never let it do things on its own, require a plan and reason. Make it evaluate its own work/plan.
Most issues i have stem from models beeing too eager. Restrain them and remove the “i can do this next…”behaviour.
Context is king - so proper mcp and documentation that is agent facing. I use serena as i can get lsp for yaml, markup and keep these docs like that
No, I think you do get it. That’s exactly right. Everything you described is absolutely valid.
Maybe the only piece you’re missing is that “almost right, but critically broken in subtle ways” turns out to actually be more than good enough for many people and many purposes. You’re describing the “success” state.
/s but also not /s because this is the unfortunate reality we live in now. We’re all going to eat slop and sooner or later we’re going to be forced to like it.
Maybe the only piece you’re missing is that “almost right, but critically broken in subtle ways”
Sure, but you have to note that it reaches that point in minutes. Sometimes on a task that would take humans a week. The power is not that it creates correct stuff, it’s that it creates almost correct stuff 100 times faster than human. Plus the typical machine benefits: it never gets tired, demotivated, etc.
So then the challenge becomes being able to be that human, who can review stuff extremely well and rapidly, being natural in probing the stuff LLMs tend to be wrong about. Sort of like the same challenge that every tech lead had before LLMs too, but just subtly different, because LLMs don’t exactly think like we do.
Or maybe we will be forced to switch off LLMs and start solving the bugs introduced by their usage using our minds.
As a professional software developer, I truly hope that is the case (and I plan to charge at least 10x my current rate after the AI bubble pops when I’m looking for my next job as I expect there to be a massive shortage of people skilled enough to actually deal with the nightmare spaghetti AI code bases)
Fun times ahead.
Their usual (crap) defense is:
a) you’re not paying enough, so of course it is crap
b) you’re not prompting right, you need to use detailed, precise language…
c) that is just anecdotal evidence, you need to do an actual study, yadda yadda.
d) it will improve…
(any other anyone has noticed?)
English is cheap to replicate, there is no science to prompting it’s asking the goddamn question.
If AI companies are so keen in keeping humans like dumbasses, that’s an issue on their part not my fucking English.
my experience with LLM’s and numerical computations like with MATLAB or GNU octave, has been poor. I assume its more of an issue that the data isn’t there, MATLAB has it’s own proprietary AI (which I don’t believe is trained on users code) and Octave has no AI associated on it’s end so the major LLM’s only get trained by the data it is prompted by users online or otherwise. Which is why if you prompt it to do a 3D plot, it will almost always pull something out of it’s ass.
your feeling of a “mental-fog” is my experience with AI in general, the language model explains the ideas well, but then the code editor does some obscure move that makes no fucking sense. also, because you’re not programming it and learning from your mistakes it makes you uncertain of your code. its unfortunate to see search engines are going to shit because of AI, because AI is not ready.
-
Did you have MCP tooling setup so it can get lsp feedback? This helps a lot with code quality as it’ll see warnings/hints/suggestions from the lsp
-
Unit tests. Unit tests. Unit tests. Unit tests.
I cannot stress enough how much less stupid LLMs get when they jave proper solid Unit tests to run themselves and compare expected vs actual outcomes.
Instead of reasoning out “it should do this” they can just run the damn test and find out.
They’ll iterate on it til it actually works and then you can look at it and confirm if its good or not.
I use Sonnet 4.5 / 4.6 extensively and, yes, its prone to getting the answer almost right but a wrong in the end.
But the unit tests catch this, and it corrects.
Example: I am working on my own fame engine with monogame and its about 95% vibe coded.
This transform math is almost 100% vibe coded: https://github.com/SteffenBlake/Atomic.Net/blob/main/MonoGame/Atomic.Net.MonoGame/Transform/TransformRegistry.cs
The reason its solid is because of this: https://github.com/SteffenBlake/Atomic.Net/blob/main/MonoGame/Atomic.Net.MonoGame.Tests/Transform/Integrations/TransformRegistryIntegrationTests.cs
Also vibe coded and then sanity checked by me by hand to confirm the math checks out for the tests.
And yes, it caught multiple bugs, but the agent automatically could respond to that, fix the bug, rerun the tests, and iterate til everything was solid.
Test Driven Development is huge for making agents self police their own code.
-
I only use AI for generating ok looking UI.
Anthropic says Methos will find bugs on FreeBSD, Bank system etc. What a bullshit.
It’s easier finding bugs than producing correct, readbale, maintainable code though-
Vibe coding, in the sense of telling the model to make codebase changes, then directly using the output produced, is 100% marketing bullshit that does not scale beyond toy examples.
Here’s the rub: Claude is extremely useful as an advanced autocomplete, if and only if you’re guiding it architecturally through every task it runs, and you vet + revise the output yourself between iterations. You cannot effectively pilot entirely from chat in a mature codebase, and you must compile robust documentation and instructions for Claude to know how to work with your codebase.
You also must aggressively manage information in the context window yourself and keep it clean. You mentioned going in circles trying to get the robot to correct itself: huge mistake. Rewind to before the error, and give it better instructions to steer it away from the pitfall it fell into. Same vein, you also need to reset ASAP after pushing into the >100k token mark, because the models start melting into putty soon after (yes, even the “extended” 1M-window ones).
I’m someone who has massively benefited from using modern LLMs in my work, but I’m also a massive hater at the same time: They’re just a tool, not magic, and have to be used with great care and attention to get reasonable results. You absolutely cannot delegate your thinking to them, because it will bite you, hard and fast.
For your use case (3D math), what I recommend is decomposing your end goal into a series of pure functions that you’ll string together. Once you have that list, that’s where Claude comes in. Have it stub those functions for you, then have it implement them one at a time, reviewing the output of every one before proceeding.
I use it and it works. It doesn’t give you the right result in one shot, but neither does manual coding. You iterate and prompt again and again. In the end, it saves a ton of time. Engineers are definitely going to lose their jobs because fewer people are needed. I know its tough to accept this and people will go through denial. Part of that is saying the AI code is junk. But, you’ll find it can produce junk and quickly fix it into the right solution faster than an engineer can. It sucks, but this is the new reality. The one thing that is cool once you embrace it is that you realize you can customize your favorite apps or even build anything you want from scratch.
It sucks, but this is the new reality.
Sorry mate, but you drank the AI koolaid from Sam Altman and the other tech oligarchs. The reality is that all of the major AI companies are deep in the red, OpenAI isn’t even making a profit with the 200$ subscription.
The only reason people are able to burn thousands of tokens to vibecode their apps is that they don’t have to pay the price for that, the companies are. This money will run out soon and then we will see the real cost for the bigger models.
If a subscription for Claude Code costs 500$ or even 1000$, will companies still pay for it or let actual humans do the work? We will see. I seriously doubt it, and I don’t want to depend on a subscription-based service to do my work while my skills are atrophying. Thank god my employer doesn’t force me to use AI.
Engineers are definitely going to lose their jobs
This kind of fear-mongering is what I despise most about the whole bubble.
I haven’t drank Koolaid. I’m talking from my experience using it in my professional software engineering job where I lead software projects. I’ve built things that used to take 20 weeks in 1 week with Claude. My employer does not really care about the cost of the tokens. And, when they can have one engineer do 20 weeks of work in 1 week, that to them is actually a cost savings. I already ask myself the question … Should I give this task to another engineer or just vibecode it myself?
OpenAI may not survive because they do have financial issues from overspending, but that barely matters. The company with the strongest coding LLM is Anthropic and it doesn’t sound like they’re having financial difficulty. Either way, now that it is clear what is possible, some company will succeed.They have incentives to do it.
Like I said, it will suck for some people, but its hard to deny the reality at this point.
I’ve built things that used to take 20 weeks in 1 week with Claude.
That’s ridiculous. You’ve either been a bad coder even before the AI hype or you’re simply lying. I have used these tools and they’re not that good or make you that fast - except when you’re just merging all of the proposed code blind and hope for the best. I fear for the future colleagues who will have to work with the raging dumpster fire you have created for them.
The company with the strongest coding LLM is Anthropic and it doesn’t sound like they’re having financial difficulty
Oh yes, they have the same problems OpenAI has. Just look into the vibecoding subreddits, you can see many people complaining about excessive rate limits and their models getting dumber. A healthy company wouldn’t try to put a cap on the token useage and introduce peak-hour throttling, that’s a big warning sign that they’re overspending as well.
its hard to deny the reality at this point
I only see one person here denying reality. You will be effed in a major way when your employer one day decides that the subscriptions are too expensive or tell you to limit your token useage.
I know it is a big change and will take some time to come to terms with it. But, it is here. I’m not going to argue anymore. It’s pointless.
Did you just pull a random infographic out of your ass without even mentioning the source? I reverse-searched it and it comes from Anthropic, of all places - the guys that run Claude Code.
Forbes took a look at that study, I love this money quote from it:
These flaws turn Anthropic’s dataset into an overstated labor-market conclusion. The study’s findings do not have the level of reliability required to sustain the breadth of the headline framing, because each conclusion rests on an exposure measure whose scope (1), construction (2, 3, 4, 5, 7), and interpretation (6, 8, 9, 10) remain contested.
So yeah, an AI company telling us that AI will theoretically replace our jobs, based on their own study with flawed data - damn, that’s trustworthy! /s
I’m not going to argue anymore. It’s pointless.
At least on this point we agree.
Removed by mod
The solutions it generated were almost write every time
Did you vibe code this post? 😂
I tried using the Sonnet 4.6 free model to convert some bash scripts to docker compose files, and it made several mistakes with case-sensitivity and failure to properly encapsulate certain path declarations that had spaces in them. if it could make such incredibly simple mistakes in converting a script to a markup language, I wouldn’t dare trust it to actually compose anything in an actual programming language like Python or Rust or C# or Swift whatever you’re using.
regress with old bugs
Have it write a test suite that enforces the correct behavior, and tell it that the test suite must pass after any change. Make sure it’s not cheating (return true) inside the test suite.
You’re probably done with this. But if you give claude a test case or two (or have it try to make them) you can have claude run the test case, and then it will iterate.
Also, aggressively use plan mode and if claude screws up more than three times do /clear, explain that it’s screwing up to it and then give it new instructions.
you need to fully be able to program to work with these things, in my experience.
you have to explain what you want very specifically, in precise programming terms.i tried a preview of chatgpt codex and it’s working better than my free version of claude, but codex creates a whole virtual programming environment, you have to connect it to a github repository, then it spins up an instance with tools you include and actually tests the code and fixes bugs before sending it back to you.
but you still need to be able to find the bugs and fix them yourself.oh and i think they work best with python, but i’ve also used ruby and dart and it’s decent.
it’s kinda like a power tool, it’ll definitely help you a lot to fix a car but if you can’t do it with wrenches it won’t help very much.I’ve never been able to program in anything more complex than BASIC and command line batch files, but I’m able to get useful output from Claude.
I’m an IT Infrastructure Manager by trade, and I got there through 20 years of supporting everything from desktop to datacenter including weird use cases like controlling systems in a research lab. On top of that, I’ve gotten under the hood of software in the form of running game servers in my spare time.
What you need to get good programs out of AI boils down to 3 things:
- The ability to teach an entity whose mistakes resemble those of a gifted child where it went wrong a step or ten back from where it’s currently looking.
- The ability to provide useful beta test / debug output regarding programs which aren’t behaving as expected. This does include looking at an error log and having some idea what that error means.
- Comfort using (either executing or compiling depending on the language) source code associated with the language you’re doing things in. This might be as simple as “How do I run a Powershell script or verify that I meet the version and module requirements for the script in question?”, or it might be as complicated as building an executable in Visual Studio. Either way whatever the pipeline is from source to execution, it must be a pipeline you’re comfortable working with. If you’re doing things anywhere outside the IT administration space, it’s reasonable to be looking at Python as the best first path rather than Powershell. Personally, I must go where supported first party modules exist for the types of work I’m developing around. In IT Administration, that’s Powershell.
I’ve made tools which automate and improve my entire department’s approach to user data, device data, application inventory, patch management, vulnerability management, and these are changes I started making with a free product three months ago, and two months back I switched to the paid version.
Programming is sort of like conversation in an alien language. For that reason, if you can give precise instructions sometimes you really can pull something new into existence using LLM coding. It’s the same reason that you could say words which have never been said in that specific order before, and have an LLM translate them to Portuguese.
I always used to talk about how everything in a computer was math, and that what interested me more than quantum computing would be a machine which starts performing the same sorts of operations on words or concepts that computers of that day ('90s and '00s when “quantum” was being slapped on everything to mean “fast” or “powerful”) were doing on math. I said that the best indicator when linguistic computing arrives would be that without ever learning to program, I’d start being able to program. I was looking at “Dragon Naturally Speaking” when I had this idea. It was one of the earliest effective speech to text programs. I stopped learning to program immediately and focused exclusively on learning operations from that point forward.
I’ve been testing the code generation abilities of LLMs for about three years. Within the last six months I feel like I’m starting to see evidence that the associations being made internally by LLMs are complex enough to begin considering them the fulfillment of my childhood dream of a “word computer”.
All the shitty stuff about environment and theft of art is all there too, which sucks, but more because our economic model sucks than because LLMs either do or do not suck. If we had a framework for meeting everybody’s basic needs, this software in its current state has the potential to turn everyone with a passion for grammatical and technical precision into a concept based developer practically overnight.
I have no qualifications to judge the quality of the generated results, yet the generated results are always of great quality.
Do you seriously not realize how out of touch this sounds?
Of course it sounds out of touch. I didn’t say it, or anything like it. Just like the other commenter, you seem to have stopped after the first sentence.
20 years of IT experience from a support perspective does qualify me to put anybody in the programming space on notice. The tools might not be as good as a talented and well trained dev, but they’re already better than a lazy dev. The output I get from Claude Code takes effort to get running. It just takes less of it than the output from my outsourced offshore MSP.
In my experience there are three ways to be successful with this tool:
- write something that already exists so it doesn’t need to think
- do all the thinking for it upfront (hello waterfall development)
- work in very small iterations that doesn’t require any leaps of logic. Don’t reprompt when it gets something wrong, instead reshape the code so it can only get it right
The issue with debugging is that it doesn’t actually think. LLMs pattern match to a chain of thought based on signals, not reasoning. For it to debug you need good signals in your code that explicitly tell what it is doing and the LLMs do not write code with that level of observability by default.
Edit: one of my workflows that I had success with is as follows:
- write a gherkin feature file describing desired functionality, maybe have the LLM create multiple scenarios after I defined one to copy from
- tell the LLM to write tests using those feature files, does an okay job but needs help making tests run in parallel.
- if the feature is simple, ask the LLM to make a plan and review it
- if the feature is complex then stub out the implementation in code and add TODOs, then direct the LLM to plan. Giving explicit goals in the code itself reduces token consumption and yield better plans
write something that already exists so it doesn’t need to think
If something already exists, it shouldn’t need to be rewritten.
Doing otherwise is a sign that something has gone wrong.
That was the case before LLMs and it is still the case today.
What they mean is rewrite something that has a LICENSE my company can’t use.
If the rewrite is based on something which has a license that your company can’t use, then the rewrite likely can’t be used either
I’m pretty sure if code is AI generated it’s likely considered original, but I’m not a lawyer by any stretch.
Only something created by a human can be copyrightable. (See the copyright status of monkey who took a selfie for precedent).
Any code written by an LLM is not copywritable because a human did not write it.
Also the company that trained the LLM is likely in breach of the licenses the code palls under.
