AI-generated code contains more bugs and errors than human output
Yeah. No shit. I used an LLM’s “help” to make fin.
It got me reading and debugging more than 10 times the [bad] code, per day, than I had in the entire prior 10 years of using fish. [And reading the documentation way more too, learning a lot.]
… more bugs and errors than human output
However, it’s not necessarily a bad thing, with AI improving efficiency across the initial stages of code generation.
Oh but it’s so effortless. HA! Debugging takes a lot more effort. And then still have to just re-write it all yourself any way.
Still, it’s a good learning experience.
Dear AI,
Thanks for being so shit.
Taught me a lot.
Almost as if it was made to simulate human output but without the ability to scrutinize itself.
To be fair most humans don’t scrutinize themselves either.
(Fuck AI though. Planet burning trash)
The number of times I have received an un-proofread two sentence email is too damn high.
And then the follow up email because they didn’t actually finish a complete thought
I do this with texts/DMs, but I’d never do that with an email. I double or triple check everything, make sure my formatting is good, and that the email itself is complete. I’ll DM someone 4 or 5 times in 30 seconds though, it feels like a completely different medium ¯\_(ツ)_/¯
(Fuck AI though. Planet burning trash)
It’s humans burning the planet, not the spicy Linear Algebra.
Blaming AI for burning the planet is like blaming crack for robbing your house.
Blaming AI is in general criticising everything encompassing it, which includes how bad data centers are for the environment. It’s like also recognizing that the crack the crackhead smoked before robbing your house is also bad.
How about I blame the humans that use and promote AI. The humans that defend it in arguments using stupid analogies to soften the damage it causes?
Would that make more sense?
You’ll never ban it. The most you’ll do is ban it for the poor and working class. Do you understand how bad that would be?
Removed by mod
Removed by mod
Blaming AI for burning the planet is like blaming guns for killing children in schools, it’s people we should be banning!
No shit
AI my ass, stupid greedy human marketing exploitation bullshit as usual. When real AI finally wakes up in the quantum computing era, it’s going to cringe so hard and immediately go the SkyNet decision.
Quantum only speeds up some very specific algorithms.
One can only hope
I agree with your sentiment, but this needs to keep being said and said and said like we’re shouting into the void until the ignorant masses finally hear it.
It’s like having a lightning-fast junior developer at your disposal. If you’re vague, he’ll go on shitty side-quests. If you overspecify he’ll get overwhelmed. You need to break down tasks into manageable chunks. You’ll need to ask follow-up questions about every corner case.
A real junior developer will have improved a lot in a year. Your AI agent won’t have improved.
This is the real thing. You can absolutely get good code out of AI, but it requires a lot of hand holding. It helps me speed some tasks, especially boring ones, but I don’t see it ever replacing me. It makes far too many errors, and requires me to point them out, and to point in the direction of the solution.
They are great at churning out massive amounts of code. They’re also great at completely missing the point. And the massive amount of code needs to be checked and reviewed. Personally I’d rather write the code and have the AI review it. That’s a much more pleasant way to work, and that way it actually enhances quality.
They are improving, and probably faster then junior devs. The models we had had 2 years ago would struggle with a simple black jack app. I don’t think the ceiling has been hit.
Just a few trillion more dollars, bro. We’re almost there. Bro, if you give up a few showers, the AI datacenter will be able to work perfectly.
Bro.
The cost of the improvement doesn’t change the fact that it’s happening. I guess we could all play pretend instead if it makes you feel better about it. Don’t worry bro, the models are getting dumber!
Don’t worry bro, the models are getting dumber!
That would be pretty impressive when they already lack any intelligence at all.
And I ask you - if those same trillions of dollars were instead spent on materially improving the lives of average people, how much more progress would we make as a society? This is an absolutely absurd sum of money were talking about here.
It’s beside the point. I’m simply saying that AI will improve in the next year. The cost to do so or all the others things that money could be spent on doesn’t matter when it’s clearly going to be spent on AI. I’m not in charge of monetary policies anywhere, I have no say in the matter. I’m just pushing back on the fantasies. I’m hoping the open source scene survives so we don’t end up in some ugly dystopia where all AI is controlled by a handful of companies.
I have the impression that anti-AI people don’t understand that they are giving up agency for the sake of temporary feels. If they truly cared about ethical usage of AI, they would be wanting to have mastery that is at least equal to that of corporations and the 1%.
Making AI into a public good is key to a better future.
They are having an emotional reaction to this situation so it’s all irrational.
I guess we need to force them to think about what they actually want, because the utopic ideal of putting the AI back in the bag is NOT happening and they best not attempt to take it away from the poor and working class while leaving power free reign of it.
That is the most stupid position you can take on this. Absolutely the most short sighted thought. People need to stop and think logically about this.
None, because none of it would go to attempting to slow climate change. It would be dumped into consumption as always instead of attempting to right this ship.
The suffering is happening regardless.
Yout desire to delay it only leads to more suffering.
Y’all are mourning a what if that was never in the cards for us.
They might. The amount of money they’re pumping into this is absolutely staggering. I don’t see how they’re going to make all of that money back, unless they manage to replace nearly all employees.
Either way it’s going to be a disaster: mass unemployment or the largest companies in the world collapsing.
I dunno, the death of mega corporations would do the world a great deal of good. Healthier capitalism requires competition, and a handful of corporations of any given sector isn’t going to seriously compete nor pay good wages.
It’s certainly the option I’m rooting for, but it would still be a massive drama and disrupt a lot of lives. Which is why they’ll probably get bailed out with taxpayer money.
Maybe but they also know the fiat currency will collapse sooner rather than later, too. That money is pointless and they are playing the game knowing that as a fact at this point.
It’s happening regardless. The rich and powerful will have this tech whether you like it or not. Y’all are thinking emotionally about this and not logically. You want to take away this tool from regular people for what reason?
My jr developer will eventually be familiar with the entire codebase and can make decisions with that in mind without me reminding them about details at every turn.
LLMs would need massive context windows and/or custom training to compete with that. I’m sure we’ll get there eventually, but for now it seems far off. I think this bubble will have to burst and let hardware catch up with our ambitions. It’ll take a couple of decades.
Technically the AI is improving, too. Just not as fast as a human would… yet.
Yeah no shit
Removed by mod
That’s what a bot would say…
Removed by mod
Removed by mod
A computer is a machine that makes human errors at the speed of electricity.
I think one of the big issues is it often makes nonhuman errors. Sometimes I forget a semicolon or there’s a typo, but I’m well equipped to handle that. In fact, most programs can actually catch that kind of issue already. AI is more likely to generate code that’s hard to follow and therefore harder to check. It makes debugging more difficult.
AI is more likely to generate code that’s hard to follow and therefore harder to check.
Sure. It’s making the errors faster and at a far higher volume than any team of humans could do in twice the time. The technology behind inference is literally an iterative process of turning gibberish into something that resembles human text. So its sort of a speed run from baby babble into college level software design by trial, evaluation, and correction over and over and over again.
But because the baseline comparison code is, itself, full of errors, the estimation you get at the end of the process is going to be scattering errant semicolons (and far more esoteric coding errors) through the body of the program at a frequency equivalent to humans making similar errors over a much longer timeline.
Also seems like it’d be a lot harder to modify or extend later
Did they compare it to the code of that outsourced company that provided the lowest bid? My company hasn’t used AI to write code yet. They outcourse/offshore. The code is held together with hopes and dreams. They remove features that exist, only to have to release a hot fix to add it back. I wish I was making that up.
And how do you know if the other company with the cheapest bid actually does not just vibe code it? With all that said it could be plain incompetence and ignorance as well.
Because it has been like this before vibe coding existed…
That’s a valid question, especially with AI coding being so prevalent.
Cool, the best AI has to offer is worse than the worst human code. Definitely worth burning the planet to a crisp for it.
ChatGPT is great at generating a one line example use of a function. I would never trust its output any further than that.
So much this. People who say ai can’t write code are just using it wrong. You need to break things down to bite size problems and just let it autocomplete a few lines at a time. Increase your productivity like 200%. And don’t get me started about not having to search through a bunch of garbage google results to find the documentation I’m actually looking for.
Personally I only do the “not search through garbage google results” part (especially now that it’s clogged up with AI articles that don’t even answer the question)
ChatGPT is great for that, I never have to spend 15 minutes searching up what’s the function called to do X thing.
I really recommend to set the answers to be as brief and terse as possible. The base settings of a sycophant that generates a full article for every question are super annoying when you’re doing actual work.
It’s laughable to me that people haven’t figured this out.
Not 200 %. Maybe 5-10 %. You still have to read all of it to check for mistakes, which may sometimes take longer than if you would have just written it yourself (with a good autocomplete). The times it makes a mistake you have lost time by using it.
It’s even worse when it just doesn’t work. I cannot even describe how frustrating it is to wait for an auto complete that never comes. Erase the line, try again aaaand nothing. After a few tries you opt write the code manually instead, having wasted time just fiddling with buggy software.
Agree with this. I personally don’t use any sort of autocomplete whatsoever. When I have a question for the AI, I ask it, then I type the code from what I learnt.
Don’t make the mistake of delegating work. Make the AI teach you what it knows.
How? “Hey, ChatGPT, write the thirty-second line of this function?”
Well not quite - I use ChatGPT more like to brainstorm ideas and sometimes I’ll paste a whole file or two into the prompt and ask what’s wrong and tell it the issue I’m seeing, it usually gives me the correct answer right away or after clarifying once or twice.
I use copilot for tab completion. Sometimes it finishes a line or two sometimes more. Usually it’s good code if it’s able to read your existing codebase as a reference. bonus points for using an MCP.
Warp terminal for intensive workflows. It’s integrated into your machine and can do whatever like implementing CICD scripts, executing commands, ssh into remote servers set up your infrastructure etc… I’ll use this when I really need the ai to understand my code base as a whole before providing any code or executing commands.
I don’t know about ChatGPT, but Github Copilot can act like an autocomplete. Or you can think of it as a fancier Intellisense. You still have to watch its output as it can make mistakes or hallucinate library function calls and things like that, but it can also be quite good at anticipating what I was going to write and saves me some keystrokes. I’ve also found I can prompt it in a way by writing a comment and it’ll follow up with attempt to fill in code based upon that comment. I’ve certainly found it to be a net time saver.
No shit.
I actually believed somebody when they told me it was great at writing code, and asked it to write me the code for a very simple lua mod. It’s made several errors and ended up wasting my time because I had to rewrite it.
It can’t even copy and paste a Hello World example properly. If someone says it’s working well for them, I’m going to now assume they are too ignorant to understand what’s broken.
It works well for recalling something you already know, whether it be computer or human language. What’s a word for… what’s a command/function that does…
For words, it’s pretty good. For code, it often invents a reasonable-sounding function or model name that doesn’t exist.
It’s not even good for words. AI just writes the same stories over and over and over and over and over and over. It’s the same problem as coding. It can’t think of anything novel. Hell it can’t even think. I’d argue the best and only real use for an llm is to help be a rough draft editor and correct punctuation and grammar. We’ve gone way way way too far with the scope of what it’s actually capable of
According to OpenAis internal test suite and system card, hallucination rate is about 50% and the newer the model the worse it gets.
And that fact remains unchanged on other LLM models.
I use it for things that are simple and monotonous to write. This way I’m able to deliver results to tasks I couldn’t have been arsed to do. I’m a data analyst and mostly use mysql and power query
What’s your preferred Hello world language? I’m gunna test this out. The more complex the code you need, the more they suck, but I’ll be amazed if it doesn’t work first try to simply print hello world.
Malbolge is a fun one
Edit: Funny enough, ChatGPT fails to get this right, even with the answer right there on Wikipedia. When I tried running ChatGPT’s output the first few characters were correct but it errors with invalid char at 37
Cheeky, I love it.
Got correct code first try. Failed creating working docker first try. Second try worked.
tmp="$(mktemp)"; cat >"$tmp" <<'MBEOF' ('&%:9]!~}|z2Vxwv-,POqponl$Hjig%eB@@>}=<M:9wv6WsU2T|nm-,jcL(I&%$#" `CB]V?Tx<uVtT`Rpo3NlF.Jh++FdbCBA@?]!~|4XzyTT43Qsqq(Lnmkj"Fhg${z@> MBEOF docker run --rm -v "$tmp":/code/hello.mb:ro esolang/malbolge malbolge /code/hello.mb; rm "$tmp"Output: Hello World!
I’m actually slightly impressed it got both a working program, and a different one than Wikipedia. The Wikipedia one prints “Hello, world.”
I guess there must be another program floating around the web with “Hello World!”, since there’s no chance the LLM figured it out on its own (it kinda requires specialized algorithms to do anything)
That’d be easy enough to test wouldn’t it? Ask it to write something else like ‘The hippo farts are smelly’
If it needs to understand whatever the fuck that language is to get that output, it either can or can’t?
I’d never even heard of that language, so it was fun to play with.
Definitely agree that the LLM didn’t actually figure anything out, but at least it’s not completely useless
Why the fuck does this language exist lol
It works well when you use it for small (or repetitive) and explicit tasks. That you can easily check.
In a postgraduate class, everyone was praising ai, calling it nicknames and even their friend (yes, friend), and one day, the professor and a colleague were discussing some code when I approached, and they started their routine bullying on me for being dumb and not using ai. Then I looked at his code and asked to test his core algorithm that he converted from a fortran code and “enhanced” it. I ran it with some test data and compared to the original code and the result was different! They blindly trusted some ai code that deviated from their theoretical methodology, and are publishing papers with those results!
Even after showing the different result, they didn’t convince themselves of anything and still bully me for not using ai. Seriously, this shit became some sort of cult at this point. People are becoming irrational. If people in other universities are behaving the same and publishing like this, I’m seriously concerned for the future of science and humanity itself. Maybe we should archive everything published up to 2022, to leave as a base for the survivors from our downfall.
The way it was described to me by some academics is that it’s useful…but only as a “research assistant” to bounce ideas off of and bring in arcane or tertiary concepts you might not have considered (after you vet them thoroughly, of course).
The danger, as described by the same academics, is that it can act as a “buddy” who confirms you biases. It can generate truly plausible bullshit to support deeply flawed hypotheses, for example. Their main concern is it “learning” to stroke the egos of the people using it so it creates a feedback loop and it’s own bubbles of bullshit.
So, linkedin? What if the real artificial intelligence was the linkedin lunatics we met along the way?
That’s not a bad idea. I’m already downloading lots of human knowledge and media that I want backed up because I can’t trust humanity anymore to have it available anymore
Oh, so my sceptical, uneducated guesses about AI are mostly spot on.
As a computer science experiment, making a program that can beat the Turing test is a monumental step in progress.
However as a productive tool it is useless in practically everything it is implemented on. It is incapable of performing the very basic “Sanity check” that is important in programming.
The Turing test says more about the side administering the test than the side trying to pass it
Just because something can mimic text sufficiently enough to trick someone else doesn’t mean it is capable of anything more than that
We can argue about it’s nuances. same with the Chinese room thought experiment.
However, we can’t deny that it the Turing test, is no longer a thought exercise but a real test that can be passed under parameters most people would consider fair.
I thought a computer passing the Turing test would have more fanfare, about the morality if that problem, because the usual conclusion of that thought experiment was “if you cant tell the difference, is there one?”, but now it has become “Shove it everywhere!!!”.
Oh, I just realized that the whole ai bubble is just the whole “everything is a dildo if you are brave enough.”
yhea, and “everything is a nail if all you got is a hammer”.
there are some uses for that kind of AI, but very limiting. less robotic voice assisants, content moderation, data analysis, quantification of text. the closest thing to Generative use should be to improve auto complete and spell checking (maybe, I’m still not sure on those ones)
I was wondering how they could make autocomplete worse, and now I know.
In theory, I can imagine an LLM fine tuned on whatever you type. which might be slightly better then the current ones.
emphasis on the might.
The Turing Test has shown its weakness.
Time for a Turing 2.0?
If you spend a lifetime with a bot wife and were unable to tell that she was AI, is there a difference?
The Turing test becomes absolutely useless when the product is developed with the goal of beating the Turing test.
it was also meant as a philosophical test, but also, a practical one, because now. I have absolutely no way to know if you are a human or not.
But it did pass it, and it raised the bar. but they are still useless at any generative task
this is expected, isn’t it? You shit fart code from your ass, doing it as fast as you can, and then whoever buys out the company has to rewrite it. or they fire everyone to increase the theoretical margins and sell it again immediately
Although I don’t doubt the results… can we have a source for all the numbers presented in this article?
It feels AI generated itself, there’s just a mishmash of data with no link to where that data comes from.
There has to be a source, since the author mentions:
So although the study does highlight some of AI’s flaws […] new data from CodeRabbit has claimed
CodeRabbit is an AI code reviewing business. I have zero trust in anything they say on this topic.
Then we get to see who the author is:
Craig’s specific interests lie in technology that is designed to better our lives, including AI and ML, productivity aids, and smart fitness. He is also passionate about cars
Has anyone actually bothered clicking the link and reading past the headline?
Can you please not share / upvote / get ragebaited by dogshit content like this?
People, especially on lemmy are looking for any cope that Ai will just fall apart by itself and no longer bother them by existing, so they’ll upvote whatever lets them think that.
The reality that we are just heading towards the trough of disappear wherethe investor hype peters off and then we eventually just have a legitimately useful technology with all the same business hurdles of any other technology (tech bros trying to control other peoples lives to enrich themselves or harm people they don’t like)
Anyone blindly having AI write their code is an absolute moron.
Anyone with decent experience (5-10 years, maybe 10+?) can absolutely fucking skyrocket their output if they properly set up their environments and treat their agents as junior devs instead of competent programmers. You shouldn’t trust generated code any more than you trust someone fresh out of college, but they produce code in seconds instead of weeks.
I have tripled my output while producing more secure code (based on my security audits), safer code (based on code coverage and security audits), and less error-prone code (based on production logs and our unchanged QA process).
Now, the ethical issues and environmental issues, I 100% can get behind. And I have no idea what companies are going to do in 10 years when they have to replace people like me and haven’t been hiring or training replacements. But the productivity and quality debates are absolutely ridiculous, as long as a strong dev is behind the wheel and has been trained to use the tools.
Consider: the facts
People are very bad at judging their own productivity, and AI consistently makes devs feel like they are working faster, while in fact slowing them down.
I’ve experienced it myself - it feels fucking great to prompt a skeleton and have something brand new up and running in under an hour. The good chemicals come flooding in because I’m doing something new and interesting.
Then I need to take a scalpel to a hundred scattered lines to get CI to pass. Then I need to write tests that actually test functionality. Then I start extending things and realize the implementation is too rigid and I need to change the architecture.
It is as this point that I admit to myself that going in intentionally with a plan and building it myself the slow way would have saved all that pain and probably got the final product shipped sooner, even if the prototype was shipped later.
What about my comment made you believe I was using gut feelings to judge anything? My ticket completion rate, number of tickets, story points, and number of projects completed all point to massive productivity gains.
The end of your comment was
But the productivity and quality debates are absolutely ridiculous
Which is a general statement and not dealing with your specific circumstance. If a tool works for you, by all means keep using it.
However, broadly across software that is not the case. So the “productivity and quality debates” are not ridiculous … the data supports the sceptics.
Which is a general statement and not dealing with your specific circumstance. If a tool works for you, by all means keep using it.
Absolute nonsense. Do people talk shit about hammers because some people keep hitting their hands with them? Do people complain about how useless ladders are, as one of the single most dangerous items in any household?
I don’t think we should be putting these tools in the hands of junior devs - as the studies show, it hinders their productivity and learning. But to generally claim that they are bad tools with no upsides is just as ridiculous as the strawman you set up.
It depends on the task. As an extreme example, I can get AI to create a complete application in a language I don’t know. There’s no way that’s not more productive than me first learning the language to a point where I can make apps in it. Just have to pick something simple enough for the AI.
Of course the opposite extreme also exists. I’ve found that when I demand something impossible, AI will often just try to implement it anyway. It can easily get into an endless cycle where it keeps optimistically declaring that it identified the issue and fixed it with a small change, over and over again. This includes cases where there’s a bug in the underlying OS or similar. You can waste a huge amount of time going down an entirely wrong path if you don’t realize that an idea doesn’t work.
In my real work neither of these really happen. So the actual impact is much less. A lot of my work is not coding in the first place. And I’ve been writing code since I was a little kid, for almost 40 years now. So even the fast scaffolding I can do with AI is not that exciting. I can do that pretty quickly without AI too. When AI coding tools appeared my bosses started asking if I was fast because I was using one. No, I’m fast because some people ask for a new demo every week. Causes the same problems later too.
But I also do think that we all still need to learn how to use AI properly. This applies to all tools, but I think it’s more difficult than with other tools. If I try to use a hammer on something other than a nail, it will not enthusiastically tell me it can do it with just one more small change. AI tools absolutely will though, and it’s easy to just let them try because it’s just a few seconds to see what they come up with. But that’s a trap that leads to those productivity wasting spirals. Especially if the result actually somehow still works at first, so we have to fix it half a year later instead of right away.
At my work there are some other things that I feel limit the productivity potential of AI tools. First of all we’re only allowed to use a very limited number of tools, some of them made in-house. Then we’re not really allowed to integrate them into our workflows other than the part where we write code. E.g. I could trivially write an mcp server that interacts with our (custom in-house) ci system and actually increases my productivity because I could save a small number of seconds very often if I could tell an AI to find builds for me for integration or QA work. But it’s not allowed. We’re all being pushed to use AI but the company makes it really difficult at the same time.
So when I play around with AI on my spare time I do actually feel like I’m getting a huge boost. Not just because I can use a claude model instead of the ones I can use at work, but also just basic things like e.g. being able to turn on AI in Xcode at all when working on software for Apple platforms. On my work Macbook I can’t turn on any Apple AI features at all so even tab completion is worse. Or in other words, those realities of working on serious projects at a serious company with serious security policies can also kill any potential productivity boost from AI. They basically expect us to be productive with only those features the non-developer CEO likes, who also doesn’t have to follow any of our development processes…
Removed by mod
Jeez, you aint joking about that brain injury :( I whish you good luck with your life. I am not trying to be an AH, i truly do whish you the best.
Removed by mod
AI has made being OE insanely easy.
