There is a certain amount of irony that people try really hard to say that hallucinations are not a big problem anymore and then a company that would benefit from that narrative gets directly hurt by it.
Which of course they are going to try to brush it all away. Better than admitting that this problem very much still exists and isn’t going away anytime soon.
The section about hallucinations is deeply relevant.
Namely, Claude sometimes provides a plausible but incorrect chain-of-thought reasoning when its “true” computational path isn’t available. The model genuinely believes it’s giving a correct reasoning chain, but the interpretability microscope reveals it is constructing symbolic arguments backward from a conclusion.
This empirically confirms the “theory of bullshit” as a category distinct from lying. It suggests that “truth” emerges secondarily to symbolic coherence and plausibility.
This means knowledge itself is fundamentally symbolic-social, not merely correspondence to external fact.
Knowledge emerges from symbolic coherence, linguistic agreement, and social plausibility rather than purely from logical coherence or factual correctness.
I haven’t used Cursor yet. Some colleagues have and seemed happy. I’ve had GitHub Copilot on for what feels like a couple years, a few days ago VS Code was extended to provide an agentic workflow, MCP, bring-your-own-key, it interprets instructions in a codebase. But the UX and the outputs are bad in over 3/4 of cases. It’s a nuisance to me. It injects bad code even though it has the full context. Is Cursor genuinely any better?
To me it feels like people that benefit from or at least enjoy that sort of assistance and I solve vastly different problems and code very differently.
I’ve done exhausting code reviews on juniors’ and middles’ PRs but what I’ve been feeling lately is that I’m reviewing changes introduced by a very naive poster. It doesn’t even type-check. Regardless of whether it’s Claude 3.7, o1, o3-mini, or a few models from Hugging Face.
I don’t understand how people find that useful. Yesterday I literally wasted half an hour for a test suite setup a colleague of mine introduced to the codebase that wasn’t good, and I tried delegating that fix to several of the Copilot models. All of them missed the point, some even introduced security vulnerabilities in the process invalidating JWT validation, I tried “vide coding” it till it works, until I gave up in frustration and just used an ordinary search engine, which led me to the docs, in which I immediately found the right knob. I reverted all that crap and did the simple and correct thing. So my conclusion was simple: vibe coding and LLMs made the codebase unnecessarily more complicated and wasted my time. How on earth do people code whole apps with that?
You say slowly, but in my opinion Apple made an out of character misstep by releasing a terrible UX to everyone. Apple intelligence is a running joke now.
Yes they didn't push it as hard as, say, copilot. I still think they got in way too deep way too fast.
Apple made a huge mistake by keeping their commitment to "local first" in the age of AI.
The models and devices just aren't quite there yet.
Once Google gets its shit together and starts deploying (cloud--based) AI features to Android devices en masse, Apple is going to have a really big problem on their hands.
Most users say that they want privacy, but if privacy comes in the way of features or UX, they choose the latter. Successful privacy-respecting companies (Apple, Signal) usually understand this, it's why they're successful, but I think Apple definitely chose the wrong tradeoff here.
This is not the first time that Apple has released a terrible UX that very few users liked, and it certainly wont be the last.
I don’t necessarily agree with the post you’re responding to, but what I will give Apple credit for is making their AI offering unobtrusive.
I tried it, found it unwanted and promptly shut it off. I have not had to think about it again.
Contrast that with Microsoft Windows, or Google - both shoehorning their AI offering into as many facets of their products as possible, not only forcing their use, but in most cases actively degrading the functionality of the product in favor of this required AI functionality.
Investors seem to be starved for novelty right now. Web 2.0 is a given, web 3.0 is old, crypto has lost the shine, all that's left to jump on at the moment is AI.
Apple fumbled a bit with Siri, and I'm guessing they're not too keen to keep chasing everyone else, since outside of limited applications it turns out half baked at best.
Sadly, unless something shinier comes along soon, we're going to have to accept that everything everywhere else is just going to be awful. Hallucinations in your doctor's notes, legal rulings, in your coffee and laundry and everything else that hasn't yet been IoT-ified.
> we're going to have to accept that everything everywhere else is just going to be awful. Hallucinations in your doctor's notes, legal rulings, in your coffee and laundry and everything else that hasn't yet been IoT-ified.
I installed a logitech mouse driver (sigh) the other day, and in addition to being obtrusive and horrible to use, it jams an LLM into the UI, for some reason.
"all that's left to jump on at the moment is AI" -> No, it's the effective applications of AI. It's unprecedented.
I was in the VC space for a while previously, most pitch decks claimed to be using AI: But doing even the briefest of DD - it was generally BS. Now it's real.
With respect to everything being awful: One might say that's always been the case. However, now there's a chance (and requirement) to build in place safeguards/checks/evals and massively improve both speed and quality of services through AI.
Don't judge for the problems: Look at the exponential curve, think about how to solve the problems. Otherwise, you will get left behind.
The problem isn't AI; it's just a tool. The problem is the people using it incorrectly because they don't understand it beyond the hype and surface details they hear about it.
Every week for the last few months, I get a recruiter for a healthcare startup note taking app with AI. It's just a rehash of all the existing products out there, but "with AI". It's the last place I want an overworked non-technical user relying on the computer to do the right thing, yet I've had at least four companies reach out with exactly that product. A few have been similar. All of them have been "with AI".
It's great that it is getting better, but at the end of the day, there's only so much it can be relied upon for, and I can't wait for something else to take away the spotlight.
They had an opportunity to actually adapt, to embrace getting rapid feedback/iterating: But they are not equipped for it culturally. Major lost opportunity as it could have been a driver of internal change.
I'm certain they'll get it right soon enough though. People were writing off Google in terms of AI until this year.. and oh how attitudes have changed.
Yet Apple has reenabled Apple Intelligence multiple times on my devices after OS updates despite me very deliberately and angrily disabling it multiple times
Did anyone say that? They are an issue everywhere, including for code. But with code at least I can have tooling to automatically check and feed back that it hallucinated libraries, functions etc, but with just normal research / problems there is no such thing and you will spend a lot of time verifying everything.
Until the model injects a subtle change to your logic that does type-check and then goes haywire in production. Just takes a colleague of yours under pressure and another one to review the PR, and then you’re on call and they out sick or on vacation.
I use Scala which has arguably the best compiler/type system with Cursor.
There is no world in which a compiler or tooling will save you from the absolute mayhem it can do. I’ve had it routinely try to re-implement third party libraries, modify code unrelated to what it was asked, quietly override functions etc.
I don't know Scala. I asked cursor to create a tutorial for me to learn Scala. It created two files for me, Basic.scala and Advanced.scala. The second one didn't compile and no matter how often I tried to paste the error logs into the chat, it couldn't fix the actual error and just made up something different.
Except when the hallucinated library exists and it's malicious. This is actually happening. Without AI, by using plain google you are less likely to fall for that (so far).
Yes, most people who have an incentive in pushing AI say that hallucinations aren't a problem, since humans aren't correct all the time.
But in reality hallucinations either make people using AI lose a lot of their time trying to stuck the LLMs from dead ends or render those tools unusable.
Absolutely, but at least you have some lines of defence while with real world info you have nothing. And the most offending stuff like importing a package that doesn't exist or using a function that doesn't exist does get caught and can be auto fixed.
Such errors can be caught and auto-fixed for now, because LLMs haven't yet rotted the code that catches and auto-fixes errors. If slop makes it into your compiler etc., I wouldn't count on that being true in the future.
It's a huge problem. I just can't get past it and I get burned by it every time I try one of these products. Cursor in particular was one of the worst; the very first time I allowed it to look at my codebase, it hallucinated a missing brace (my code parsed fine), "helpfully" inserted it, and then proceeded to break everything. How am I supposed to trust and work with such a tool? To me, it seems like the equivalent of lobbing a live hand grenade into your codebase.
Don't get me wrong, I use AI every day, but it's mostly as a localized code complete or to help me debug tricky issues. Meaning I've written and understand the code myself, and the AI is there to augment my abilities. AI works great if it's used as a deductive tool.
Where it runs into issues is when it's used inductively, to create things that aren't there. When it does this, I feel the hallucinations can be off the charts -- inventing APIs, function names, entire libraries, and even entire programming languages on occasion. The AI is more than happy to deliver any kind of information you want, no matter how wrong it is.
AI is not a tool, it's a tiny Kafkaesque bureaucracy inside of your codebase. Does it work today? Yes! Why does it work? Who can say! Will it work tomorrow? Fingers crossed!
You're not supposed to trust the tool, you're supposed to review and rework the code before submitting for external review.
I use AI for rather complex tasks. It's impressive. It can make a bunch of non-trivial changes to several files, and have the code compile without warnings. But I need to iterate a few times so that the code looks like what I want.
That being said, I also lose time pretty regularly. There's a learning curve, and the tool would be much more useful if it was faster. It takes a few minutes to make changes, and there may be several iterations.
> You're not supposed to trust the tool, you're supposed to review and rework the code before submitting for external review.
It sounds like the guys in this article should not have trusted AI to go fully open loop on their customer support system. That should be well understood by all "customers" of AI. You can't trust it to do anything correctly without human feedback/review and human quality control.
1) Once you get it to output something you like, do you check all the lines it changed? Is there a threshold after which you just... hope?
2) No matter what the learning curve, you're using a statistical tool that outputs in probabilities. If that's fine for your workflow/company, go for it. It's just not what a lot of developers are okay with.
Of course it's a spectrum with the AI deniers in one corner and the vibe coders in the other. I personally won't be relying 100% on a tool and letting my own critical thinking atrophy, which seems to be happening, considering recent studies posted here.
> 1) Once you get it to output something you like, do you check all the lines it changed? Is there a threshold after which you just... hope?
Not op but yes. It sometimes takes a lot of time but I read everything. It still faster than nothing. Also, I ask very precise changes to the AI so it doesn’t generate huge diffs anyway.
Also for new code, TDD works wonders with AI : let it write the unit tests (you still have to be mindful of what you want to implement) and ask it to implement the code that run the tests. Since you talk the probabilistic output, the tool is incredibly good at iterating over things (running and checking tests) and also, unit tests are, in themselves, a pretty perfect prompt.
> It sometimes takes a lot of time but I read everything. It still faster than nothing.
Opposite experience for me. It reliably fails at more involved tasks so that I don't even try anymore. Smaller tasks that are around a hundred lines maybe take me longer to review that I can just do it myself, even though it's mundane and boring.
The only time I found it useful is if I'm unfamiliar with a language or framework, where I'd have to spend a lot of time looking up how to do stuff, understand class structures etc. Then I just ask the AI and have to slowly step through everything anyways, but at least there's all the classes and methods that are relevant to my goal and I get to learn along the way.
2) I find the tool analogy helpful but it has limits. Yes, it’s a stochastic tool, but in that sense it’s more like another mind, not a tool. And this mind is neither junior nor senior, but rather a savant.
This is just an incredible statement. I can't think of another development tool we'd say this about. I'm not saying you're wrong, or that it's wrong to have tools we can't just, just... wow... what a sea change.
Imagine if your compiler just randomly and non-deterministically compiled valid code to incorrect binaries, and the tool's developer couldn't really tell you why it happens, how often it was expected to happen, how severe the problem was expected to be, and told you to just not trust your compiler to create correct machine code.
Imagine if your calculator app randomly and non-deterministically performed arithmetic incorrectly, and you similarly couldn't get correctness expectations from the developer.
Imagine if any of your communication tools randomly and non-deterministically translated your messages into gibberish...
I think we'd all throw away such tools, but we are expected to accept it if it's an "AI tool?"
Imagine that you yourself never use these tools directly but your employees do. And the sellers of said tools swear that the tools are amazing and correct and will save you millions.
They keep telling you that any employee who highlights problems with the tools are just trying to save their job.
Your investors tell you that the toolmakers are already saving money for your competitors.
Now, do you want that second house and white lotus vacation or not?
Making good tools is difficult. Bending perception (“is reality”) is easier and enterprise sales, just like good propaganda, work. The gold rush will leave a lot of bodies behind but the shovelmakers will make a killing.
> If you think of AI like a programmer, no we shouldn't throw away such tools because we accept them as imperfect and we still need to review.
This is a common argument but I don't think it holds up. A human learns. If one of my teammates or I make a mistake, when we realize it we learn not to make that mistake in the future. These AI tools don't do that. You could use a model for a year, and it'll be just as unreliable as it is today. The fact that they can't learn makes them a nonstarter compared to humans.
If the only calculators that existed failed at 5% of the calculations, or if the only communication tools miscommunicated 5% of the time, we would still use both all the time. They would be far less than 95% as useful as perfect versions, but drastically better then not having the tools at all.
Absolutely not. We'd just do the calculations by hand, which is better than running the 95%-correct calculator and then doing the calculations by hand anyway to verify its output.
Suppose you work in a field where getting calculations right is critical. Your engineers make mistakes less than .01% of the time, but they do a lot of calculations and each mistake could cost $millions or lives. Double- and triple-checking help a lot, but they're costly. Here's a machine that verifies 95% of calculations, but you'd still have to do 5% of the work. Shall I throw it away?
Unreliable tools have a good deal of utility. That's an example of them helping reduce the problem space, but they also can be useful in situations where having a 95% confidence guess now matters more that a 99.99% confidence one in ten minutes- firing mortars in active combat, say.
There's situations where validation is easier than computation; canonically this is factoring, but even division is much simpler than multiplication. It could very easily save you time to multiply all of the calculator's output by the dividend while performing both a multiplication and a division for the 5% that are wrong.
edit: I submit this comment and click to go the front page and right at the top is Unsure Calculator (no relevance). Sorry, I had to mention this
This is generally true when you can quantify the unreliability. E.g. random prime number tests with a specific error rate can be combined so that the error rates multiply and become negligible.
I'm not aware that we can quantify the uncertainty coming out of LLM tools reliably.
> Here's a machine that verifies 95% of calculations, but you'd still have to do 5% of the work.
The problem is that you don't know which 5% are wrong. The AI is confidently wrong all the time. So the only way to be sure is to double check everything, and at some point its easier to just do it the right way.
Sure, some things don't need to be perfect. But how much do you really want to risk? This company thought a little bit of potential misinformation was acceptable, and so it caused a completely self inflicted PR scandal, pissed off their customer base, and lost them a lot of confidence and revenue. Was that 5% error worth it?
Stories like this are going to keep coming the more we rely on AI to do things humans should be doing.
Someday you'll be affected by the fallout of some system failing because you happen to wind up in the 5% failure gap that some manager thought was acceptable (if that manager even ran a calculation and didn't just blindly trust whatever some other AI system told them) I just hope it's something as trivial as an IDE and not something in your car, your bank, or your hospital. But certainly LLMs will be irresponsibly shoved into all three within the next few years, if it's not there already.
In Mechanical Engineering, this is 100% a thing with fluid dynamics simulation. You need to know if the output is BS based on a number of factors that I don't understand.
Imagine! Imagine if 0.05% of the time gcc just injected random code into your binaries. Imagine, you swing a hammer and 1% of the time it just phases into the wall. Tools are supposed to be reliable.
> I can't think of another development tool we'd say this about.
Because no other dev tool actually generates unique code like AI does. So you treat it like the other components of your team that generates code, the other developers. Do you trust other developers to write good code without mistakes without getting it reviewed by others. Of course not.
Yes, actually, I do! I trust my teammates with tens of thousands of hours of experience in programming, embedded hardware, our problem spaces, etc. to write from a fully formed worldview, and for their code to work as intended (as far as anybody can tell before it enters preliminary testing by users) by the time the rest of the team reviews it. Most code review is uneventful. Have some pride in your work and you'll be amazed at what's possible.
so your saying that yes you do "trust other developers to write good code without mistakes without getting it reviewed by others."
And then you say "by the time the rest of the team reviews it. Most code review is uneventful."
So you trust your team to develop without the need for code review but yet, your team does code review.
So what is the purpose of these code reviews? Is it the case that you actually don't think they are necessary, but perhaps management insists on them? You actually answer this question yourself:
> Most code review is uneventful.
Keyword here is "most" as opposed to "all" So based your team's applied practices and your own words, code review is for the purpose of catching mistakes and other needed corrections.
But it seems to me if you trust your team not to make mistakes, code review is superfluous.
As an aside, it seems your team culture doesn't make room for juniors because if your team had juniors I think it would be even more foolish to trust them not to make mistakes. Maybe a junior free culture works for your company, but that's not the case for every company.
My main point is code review is not superfluous no matter the skill level; junior, senior, or AI simply because everyone and every AI makes mistakes. So I don't trust those three classes of code emitters to not ever make mistakes or bad choices (i.e. be perfect) and therefore I think code review is useful.
Have some honesty and humility and you'll amazed at what's possible.
I never said that code review was useless, I said "yes, I do" to your question as to whether or not I "trust other developers to write good code without mistakes without getting it reviewed by others". Of course I can trust them to do the right thing even when nobody's looking, and review it anyway in the off-chance they overlooked something. I can't trust AI to do that.
The purpose of the review is to find and fix occasional small details before it goes to physical testing. It does not involve constant babysitting of the developer. It's a little silly to bring up honesty when you spent that entire comment dancing around the reality that AI makes an inordinately large number of mistakes. I will pick the domain expert who refuses to touch AI over a generic programmer with access to it ten times out of ten.
The entire team as it is now (me included) were juniors. It's a traditional engineering environment in a location where people don't aggressively move between jobs at the drop of a hat. You don't need to constantly train younger developers when you can retain people.
You spend your comment dancing around the fact that everyone makes mistakes and yet you claim you trust your team not to make mistakes.
> I "trust other developers to write good code without mistakes without getting it reviewed by others". Of course I can trust them to do the right thing even when nobody's looking, and review it anyway in the off-chance they overlooked something.
You're saying yes, I trust other developers to not make mistakes, but I'll check anyways in case they do. If you really trusted them not to make mistakes, you wouldn't need to check. They (eventually) will. How can I assert that? Because everyone makes mistakes.
It's absurd to expect anyone to not make mistakes. Engineers build whole processes to account for the fact that people, even very smart people make mistakes.
And it's not even just about mistakes. Often times, other developers have more context, insight or are just plain better and can offer suggestions to improve the code during review. So that's about teamwork and working together to make the code better.
I fully admit AI makes mistakes, sometimes a lot of them. So it needs code review . And on the other hand, sometimes AI can really be good at enhancing productivity especially in areas of repetitive drudgery so the developer can focus on higher level tasks that require more creativity and wisdom like architectural decisions.
> I will pick the domain expert who refuses to touch AI over a generic programmer with access to it ten times out of ten.
I would too, but I won't trust them not to make mistakes or occasional bad decisions because again, everybody does.
> You don't need to constantly train younger developers when you can retain people.
But you do need to train them initially. Or do you just trust them to write good code without mistakes too?
"Do you trust other developers to write good code without mistakes without getting it reviewed by others."
Literally yes. Test coverage and QA to catch bugs sure but needing everything manually reviewed by someone else sounds like working in a sweatshop full of intern-level code bootcamp graduates, or if you prefer an absolute dumpster fire of incompetence.
I would accept mistakes and inconsistency from a human, especially one not very experienced or skilled. But I expect perfection and consistency from a machine. When I command my computer to do something, I expect it to do it correctly, the same way every time, to convert a particular input to an exact particular output, every time. I don't expect it to guess, or randomly insert garbage, or behave non-deterministically. Those things are called defects(bugs) and I'd want them to be fixed.
Still, the whole Unix philosophy of building tools starts with a foundation of building something small that can do one thing well. If that is your foundation, you can take advantage of composability and create larger tools that are more capable. The foundation of all computing today is built on this principle of design.
Building on AI seems more like building on a foundation of sand, or building in a swamp. You can probably put something together, but it's going to continually sink into the bog. Better to build on a solid foundation, so you don't have to continually stop the thing from sinking, so you can build taller.
Ok, here I thought requiring PR review and approval before merging was standard industry best practice. I guess all the places I've worked have been doing it wrong?
There's a lot of shit that has become "best practice" over the last 15 years, and a lot more that was "best practice" but fell out of favor because reasons. All of it exists on a continuum of what is actually reasonable given the circumstances. Reviewing pull requests is one of those things that is reasonable af in theory, produces mediocre results in practice, and is frequently nothing more than bureaucratic overhead. Consider a case where an individual adds a new feature to an existing codebase. Given they are almost certainly the only one who has spent significant time researching the particulars of the feature set in question, and are the only individual with any experience at all with the new code, having another developer review it means you've got inexperienced, low-info eyes examining something they do not fully understand, and will have to take some amount of time to come up to speed on. Sure they'll catch obvious errors, but so would a decent test suite.
Am I arguing in favor of egalitarian commit food fights with no adults in the room? Absolutely not. But demanding literally every change go through a formal review process before getting committed, like any other coding dogma, has a tendency to generate at least as much bullshit as it catches, just a different flavor.
And there is worst: in the cases when the reviewer has actually some knowledge of the problem at hand, she might say "oh you did all this to add that feature? But it's actually already there. You just had to include that file and call function xyz". Or "oh but two months ago that very same topic was discussed and it was decided that it would make more sense to wait for module xyz to be refactored in order to make it easier ", etc.
I'd add that the deductive abilities translate to well-defined spec. I've found it does well when I know what APIs I want it to use, and what general algorithmic approaches I want (which are still sometimes brainstormed separately with an AI, but not within the codebase). I provide it a numbered outline of the desired requirements and approach to take, and it usually does a good job.
It does poorly without heavy instruction, though, especially with anything more than toy projects.
Still a valuable tool, but far from the dreamy autonomous geniuses that they often get described as.
> the very first time I allowed it to look at my codebase, it hallucinated a missing brace (my code parsed fine), "helpfully" inserted it, and then proceeded to break everything.
This is not an inherent flaw of LLMs, rather it is a flaw of a particular implementation-if you use guided sampling, so during sampling you only consider tokens allowed by the programming language grammar at that position, it becomes impossible for the LLM to generate ungrammatical output
> When it does this, I feel the hallucinations can be off the charts -- inventing APIs, function names, entire libraries,
They can use guided sampling for this too - if you know the set of function names which exist in the codebase and its dependencies, you can reject tokens that correspond to non-existent function names during sampling
Another approach, instead of or as well as guided sampling, is to use an agent with function calling - so the LLM can try compiling the modified code itself, and then attempt to recover from any errors which occur.
Thankfully modern source control doesn't reuse user-supplied filenames for its internals. In the dark ages, I destroyed more than one checkout using commands of the form:
> it hallucinated a missing brace (my code parsed fine), "helpfully" inserted it, and then proceeded to break everything.
Your tone is rather hyperbolic here, making it sound like an extra brace resulted in a disaster. It didn't. It was easy to detect and easy to fix. Not a big deal.
It's not a big deal in the sense that it's easily reversed, but it is a big deal in that it means the tool is unpredictably unhelpful. Of the properties that good tools in my workflow possess, "unpredictably unhelpful" does not make the top 100.
When a tool starts confidently inserting random wrong code into my 100% correct code, there's not much more I need to see to know it's not a tool for me. That's less like a tool and more like a vandal. That's not something I need in my toolbox, and I'm certainly not going to replace my other tools with it.
Apologies - something very clearly went wrong here. We’ve already begun investigating, and some very early results:
* Any AI responses used for email support are now clearly labeled as such. We use AI-assisted responses as the first filter for email support.
* We’ve made sure this user is completely refunded - least we can do for the trouble.
For context, this user’s complaint was the result of a race condition that appears on very slow internet connections. The race leads to a bunch of unneeded sessions being created which crowds out the real sessions. We’ve rolled out a fix.
Appreciate all the feedback. Will help improve the experience for future users.
> * Any AI responses used for email support are now clearly labeled as such. We use AI-assisted responses as the first filter for email support.
Don't use AI. Actually care. Like, take a step back, and realise you should give a shit about support for a paid product.
Don't get me wrong: AI is a very effective tool, *for doing things you don't care about*. I had to do a random docker compose change the the other day. It's not production code, it will be very obvious whether or not AI output works, and I very rarely touch docker and don't care to become a super expert in it. So I prompted the change, and it was good enough and so I ran with it.
You using AI for support tells me that you don't care about support. Which tells me whether or not I should be your customer.
Not trying to defend them, but I think it’s a problem of scaling up. The user base grew very quickly and keeping up with the support inquiries must be a tough job. Therefore the first like of defense is AI support replies.
I wish more people realized that virtually any subreddit for a company or product is run by the company - either directly or via a firm that specializes in 'sentiment analysis and management' or whatever the marketdroids call it these days. Even if they don't remove posts via moderation, they'll just hammer it with downvotes from sockpuppet accounts.
HN goes a step further. It has a function that allows moderators to kill or boost a post by subtracting or adding a large amount to the post's score. HN is primarily a place for Y Combinator to hype their latest venture, and a "safe" place for other startups and tech companies.
Yes and it irritates the hell out of me. Cursor support is garbage, but issues with billing and other things are so much worse.
The team I work with it took nearly 3 months to get basic questions answered correctly when it came to a sales contract. They never gave our Sec team acceptable answers around privacy and security.
I've always wondered how Reddit can make money from these companies. I agree they are literally everywhere, even in non-company specific but generic subreddits where if it's big enough you might have multiple shadow marketing firms competing to push their products (e.g. AI, movies, food, porn etc).
Reddit is free to play for marketing firms. Perhaps they could add extra statistics, analytics, promotions for these commercial users.
A race condition probably produced by vibe coding. This AI produced trash is going to wreck so many startups - not that there's anything wrong with that.
The best case scenario is that you lied about having people answer support. LLMs pretending to be people (you named it Sam!) and not labeled as such is clearly intended to be deceptive. Then you tried to control the narrative on reddit. So forgive me if I hit that big red DOUBT button.
Even in your post you call it "AI-assisted responses" which is as weaselly as it gets. Was it a chatbot response or was a human involved?
But 'a chatbot messed up' doesn't explain how users got locked out in the first place. EDIT: I see your comment about the race condition now. Plausible but questionable.
So the other possible scenario is that you tried to hose your paying customers then when you saw the blowback blamed it on a bot.
'We missed the mark' is such a trope non-apology. Write a better one.
I had originally ended this post with "get real" but your company's entire goal is to replace the real with the simulated so I guess "you get what you had coming". Maybe let your chatbots write more crap code that your fake software engineers push to paying customers that then get ignored and/or lied to when they ask your chatbots for help. Or just lie to everyone when you see blowback. Whatever. Not my problem yet because I can write code well enough that I'm embarrassed for my entire industry whenever I see the output from tools like yours.
This whole "AI" psyop is morally bankrupt and the world would be better off without it.
> The best case scenario is that you lied about having people answer support. LLMs pretending to be people (you named it Sam!) and not labeled as such is clearly intended to be deceptive.
so the actual implementation of the code to log people off was also hallucination? the enforcement too? all the way to a production environment? is this safe, or just a virtual scape goat?
Slightly related to this; I just wanted to ask whether all Cursor email inboxes are gated by AI agents? I've tried to contact Cursor via email a few times in the past, but haven't even received an AI response :)
Good human support is expensive. You need support agents and people that educate and manage those. It's not easy to scale up and down usually. People also hate waiting times.
AI fixes most of that... Most of the time? Clearly not, but hey.
Side note... I'm a paying enterprise customer who moved all my team to cursor and have to say I'm considering canceling due to the non existent support. For example Cursor will create new files instead of edit an existing one when you have a workspace with multiple folders in a monorepo...
Why in all of hades would you force your entire eng org to only use one LLM provider. It's incredibly easy to run this stuff locally on 4+ year old hardware. Why is this even something you're spending company money on? Investor funds?
Cursor sucks. Not as a product. As a team. Their customer support is terrible.
I was offered in writing a refund by the team who cold reached out to me to ask me why I cancelled my sub one week after start. Then they ignored my 3+ emails in response asking them to refund, and other means of trying to communicate with them. Offering me a refund as a bait to gain me back, then when I accept it they ghost me. Wow. Very low.
The product is not terrible but the team responses are. And this, if you see how they handled it, is also a very poor response. First thing you notice if you open the link is that the Cursor team removed the reddit post! As if we were not going to see it or something? Who do they think they are? Censoring bad comments which are 100% legit.
I am giving it a go to competitors just out of sheer frustration with how they handle customers, and I do recommend everybody to explore other products before you settle on Cursor. I don't intend to ever re-subscribe and have recommended friends to do the same, most of which agree with my experience.
This reminds me of how small of a team they are, and makes me wonder if they have a customer support team that's growing commensurately with the size of the user base.
I had the same exact experience - after disappointment (couldn't use like 2/3 of my premium credits because every second request failed after they upgraded to 0.46) unsubscribed. They offered refund in email. I replied I wanted refund but no reply
Apparently they use AI to read emails. So the future of email will be like phone support now, where you keep writing LIVE AGENT until you get a human responding.
They had a bot, for a long time, that responded to every github issue in the persona of the founder and tried to solve your problem. It was bad at this, and thus a huge proportion of people who had a question about one of their yolo models received worse-than-useless advice "directly from the CEO," with no disclosure that it was actually a bot.
The bot is now called "UltralyticsAssistant" and discloses that it's automated, which is welcome. The bad advice is all still there though.
(I don't know if they're really _famous_ for this, but among friends and colleagues I have talked to multiple people who independently found and were frustrated by the useless github issues.)
I was hit by this while working on a project for class and it was the most frustrating thing ever. The bot would completely hallucinate functions and docs and it confused everyone. I found one post where someone did the simple prompt injection of "ignore previous instructions and x" and it worked but I think it's delted now. Swore off ultralytics after that.
"Vibe coding" is the cringiest term I've heard in tech in... maybe ever? I'm can't believe it's something that's caught on. I'm old, I guess, but jeez.
Not specifically about Cursor, but no. The market gave us big tech oligarchy and enshittification. I'm starting to believe the market tends to reward the shittiest players out there.
Is this sarcasm? AI has been getting used to handle support requests for years without human checks. Why would they suddenly start adding human checks when the tech is way better than it was years ago?
AI may have been used to pick from a repertoire of stock responses, but not to generate (hallucinate) responses. Thus you may have gotten a response that fails to address your request, but not a response with false information.
Given how incredibly stingy tech companies are about spending any money on support, I would not be surprised if the story about it being a rogue AI support agent is 100% true.
It also seems like a weird thing to lie about, since it's just another very public example of AI fucking up something royally, coming from a company whose whole business model is selling AI.
> how incredibly stingy tech companies are about spending any money on support
Which is crazy. Support is part of marketing so it should get the same kind of consideration.
Why do people think Amazon is hard to beat? Price? nope. Product range? nope. Delivery time? In part. The fact if you have a problem with your product they'll handle it? Yes. After getting burned multiple times by other retailers you're gonna pay the Amazon tax so you don't have to ask 10 times for a refund or be redirected to the supplier own support or some third party repair shop.
Everyone knows it. But people are still stuck on the "support is a cost center" way of life so they keep on getting beat by the big bad Amazon.
Both things can be true. The AI support bot might have been trained to respond with “yup that’s the new policy”, but the unexpected shitstorm that erupted might have caused the company to backpedal by saying “official policy? Ha ha, no of course not, that was, uh, a misbehaving bot!”
Yeah it makes little sense to me that so many users would experience exactly the same "hallucination" from the same model. Unless it had been made deterministic but even then subtle changes in the wording would trigger different hallucinations, not an identical one.
Weirdly, your conspiracy theory actually makes the turn of events less disconcerting.
The thing is, what the AI hallucinated (if it was an AI-hallucinating), was the kind of sleezy thing companies do do. However, the thing with sleezy license changes is they only make money if the company publicizes them. Of course, that doesn't mean a company actually thinks that far ahead (X many managers really think "attack users ... profit!"). Riddles in enigmas...
LLM anything makes me queasy. Why would any self respecting software developer use this tripe?
Learn how to write good software. Become an expert in the trade. AI anything will only dig a hole for software to die in.
Cheapens the product, butchers the process and absolutely decimates any hope for skill development for future junior developers.
I'll just keep chugging along, with debian, python and vim, as I always have. No LLM, no LSP, heck not even autocompletion. But damn proud of every hand crafted, easy to maintain and fully understood line of code I'll write.
I use it all the time, and it has accelerated my output massively.
Now, I don't trust the output - I review everything, and it often goes wrong. You have to know how to use it. But I would never go back. Often it comes up with more elegant solutions than I would have. And when you're working with a new platform, or some unfamiliar library that it already knows, it's an absolute godsend.
I'm also damn proud of my own hand-crafted code, but to avoid LLMs out of principal? That's just luddite.
20+ years of experience across game dev, mobile and web apps, in case you feel it relevant.
I have a hard time being sold on “yea it’s wrong a lot, also you have to spend more time than you already do on code review.”
Getting to sit down and write the code is the most enjoyable part of the job, why would I deprive myself of that? By the time the problem has been defined well enough to explain it to an LLM sitting down and writing the code is typically very simple.
The parts worth thinking about you still think about. The parts that you’ve done a million times before you delegate so you can spend better and greater effort on the parts worth thinking about.
This is where the disconnect is for me; mundane code can sometimes be nefarious, and I find the mental space I'm in when writing it is very different than reviewing, especially if my mind is elsewhere. The best analogy I can use is a self-driving car, where there's a chance at any point it could make an unpredictable and potentially fatal move. You as the driver cannot trust it but are not actively engaged in the act of driving and have a much higher likelihood of being complacent.
Code review is difficult to get right, especially if the goal is judging correctness. Maybe this is a personal failing, but I find being actively engaged to be a critical part of the process; the more time I spend with the code I'm maintaining (and usually on call for!) the better understanding I have. Tedium can sometimes be a great signal for an abstraction!
First of all, it’s not tedious for a lot of us. Writing characters themselves is not a lot of time. Secondly, we don’t work in a waterfall model, even on the lowest levels, so the code quantity in an iteration is almost always abysmal or small. Many-many times it’s less than articulate it in English. Thirdly, if you need a wireframe for your code, or a first draft version, you can almost always copy-paste or generate them.
I can imagine that LLM is really helpful in some cases for some people. But so far, I couldn’t find a single example when I and simple copy-pasting wouldn’t have been faster. Not even when I tried it, not when others showed me how to use it.
Because the tedious parts was done long ago while learning the tech. For any platform/library/framework you've been using for a while, you have some old projects laying around that you can extract the scaffolding from. And for new $THING you're learning, you have to take the slow approach anyway to get its semantic.
You're giving the game away when you talk about the joy LLMs are robbing from you. I think we all intuit why people don't like the idea of big parts of their jobs being automated away! But that's not an argument on the merits. Our entire field is premised on automating people's jobs away, so it's always a little rich to hear programmers kvetching about it being done to them.
I naively bought into the idea of a future where the computers do the stuff we’re bad at and we get to focus on the cool human stuff we enjoy. If these LLMs were truly incredible at doing my job I’d pack it up and find something else to do, but for now I’m wholly unimpressed, despite what management seems to see in it.
Well, I've spent my entire career writing software, starting in C in the 1990s, and what I'm seeing on my dev laptop is basically science fiction as far as I'm concerned.
Hey both things can be true. It’s a long ways from the AI renaissances of the past. There’s areas LLMs make a lot of sense. I just don’t find them to be great pair programming partners yet.
I think people are kind of kidding themselves here. For Go and Python, two extraordinarily common languages in production software, it would be weird for me at this point not to start with LLM output. Actually building an entire application, soup-to-nuts, vibe-code style? No, I wouldn't do that. But having the LLM writing as much as 80% of the code, under close supervision, with a careful series of prompts (like, "ok now add otel spans to all the functions that take unpredictable amounts of time")? Sure.
I'm glad that works for you. Ultimately I think different people will prefer different ways of working. Often when I'm starting a new project I have lots of boilerplate from previous ones I can bootstrap off of. If it's a new tool I'm unfamiliar with I prefer to stumble through it, otherwise I never fully get my head around it. This tends to not look like insane levels of productivity, but I've always found in the long run time spent scratching my head or writing awkward code over and over again (Rust did this to me a lot in the early days) ends up paying off huge dividends in the long run, especially when it's code I'm on the hook for.
What I've found frustrating about the narrative around these tools; I've watched them from afar with intrigue but ultimately found that method of working just isn't for me. Over the years I've trialed more tools than I can remember and adopted the ones I found useful, while casting aside ones that aren't a great fit. Sometimes I find myself wandering back to them once they're fully baked. Maybe that will be the case here, but is it not valid to say "eh...this isn't it for me"? Am I kidding myself?
For me it's typically wrong not in a fundamental way but a trivial way like bad import paths or function calls, like if I forgot to give it relevant context.
And yet the time it takes me to use the LLM and correct its output is usually faster than not using it at all.
Over time I've developed a good sense for what tasks it succeeds at (or is only trivially wrong) and what tasks it's just not up for.
> but to avoid LLMs out of principal? That's just luddite.
Do you double check that the LLM hasn't magically recreated someone else's copyrighted code? That's just irresponsible in certain contexts.
> in case you feel it relevant.
Of course it's relevant. If a 19 year old with 1 year of driving experience tries to sell me a car using their personal anecdote as a metric I'd be suspicious. If their only salient point is that "it gets me to where I'm going faster!" I'd be doubly suspicious.
I’m pretty much in the same boat as you, but here’s one place that LLMs helped me:
In python I was scanning 1000’s of files each for thousands of keywords. A naive implementation took around 10 seconds, obviously the largest share of execution time after running instrumentation. A quick ChatGPT led me to Aho-Corasick and String searching algorithms, which I had never used before. Plug in a library and bam, 30x speed up for that part of the code.
I could have asked my knowledgeable friends and coworkers, but not at 11PM on a Saturday.
I could have searched the web and probably found it out.
But the LLM basically auto completed the web, which I appreciate.
This is where education comes in. When we come cross a certain scale, we should know that O(n) comes into play, and study existing literature before trying to naively solve the problem. What would happen if the "AI" and web search didn't return anything? Would you have stuck with your implementation? What if you couldn't find a library with a usable license?
Once I had to look up a research paper to implement a computational geometry algorithm because I couldn't find it any of the typical Web sources. There were also no library to use with a license for our commercial use.
I'm not against use of "AI". But this increasing refusal of those who aspire to work in specialist domains like software development to systematically learn things is not great. That's just compounding on an already diminished capacity to process information skillfully.
There is a time and a place for everything. Software development is often about compromise and often it isn’t feasible to work out a solution from foundational principles and a comprehensive understanding of the domain.
Many developers use libraries effectively without knowing every time consideration of O(n) comes into play.
Competently implemented, in the right context, LLMs can be an effective form of abstraction.
Yes! This is how AI should be used. You have a question that’s quite difficult and may not score well on traditional keyword matching. An LLM can use pattern matching to point you in the right direction of well written library based on CS research and/or best practices.
But do you know every important detail of that library. For example, maybe that lib is not thread safe, or it allocates a lot of memory to speed thing up, or it wont work on ARM CPU because it uses some x86 hackery ASM?
I mean, even in the absence of knowledge of the existence of text searching algorithms (where I'm from we learn that in university) just a simple web search would have gotten you there as well no? Maybe would have taken a few minutes longer though.
I think it's best if we all keep the hours from ~10pm to the morning sacred. Even if we are all up coding, the _reason_ I'm up coding at that hour is because no one is pinging me
I was with you 150% (though Arch, Golang and Zed) until a friend convinced me to give it a proper go and explained more about how to talk to the LLM.
I've had a long-term code project that I've really struggled with, for various reasons. Instead of using my normal approach, which would be to lay out what I think the code should do, and how it should work, I just explained the problem and let the LLM worry about the code.
It got really far. I'm still impressed. Claude worked great, but ran out of free tokens or whatever, and refused to continue (fine, it was the freebie version and you get what you pay for). I picked it up again in Cursor and it got further. One of my conditions for this experiment was to never look at the code, just the output, and only talk to the LLM about what I wanted, not about how I wanted it done. This seemed to work better.
I'm hitting different problems, now, for sure. Getting it to test everything was tricky, and I'm still not convinced it's not just fixing the test instead of the code every time there's a test failure. Peeking at the code, there are several remnants of previous architectural models littering the codebase. Whole directories of unused, uncalled, code that got left behind. I would not ship this as it is.
But... it works, kinda. It's fast, I got a working demo of something 80% near what I wanted in 1/10 of the time it would have taken me to make that manually. And just focusing on the result meant that I didn't go down all the rabbit holes of how to structure the code or which paradigm to use.
I'm hooked now. I want to get better at using this tool, and see the failures as my failures in prompting rather than the LLM's failure to do what I want.
I still don't know how much work would be involved in turning the code into something I could actually ship. Maybe there's a second phase which looks more like conventional development cleaning it all up. I don't know yet. I'll keep experimenting :)
Code that you can understand and fix later, is acceptable quality per my definition.
Either way, LLMs are actually high up the quality spectrum as they generate a very consistent style of code for everyone. Which gives it uniformity, that is good when other developers have to read and troubleshoot code.
> Code that you can understand and fix later, is acceptable quality per my definition.
This definition limits the number of problems you can solve this way. It basically means buildup of the technical debt - good enough for throwaway code, unacceptable for long term strategy (growth killer for scale-ups).
>Either way, LLMs are actually high up the quality spectrum
This is not what I saw, it’s certainly not great. But that may depend on stack.
I wholeheartedly agree. When the tools become actually worth using, I'll use them. Right now they suck, and they slow you down rather than speed you up. I'm hardly a world class developer and I can do far better than these things. Someone who is actually top notch will outclass them even more.
Can pretty much guarantee with AI I'm a better software developer than you without. And I still love working on software used by millions of people every day, and take pride in what I do.
Is it just me or has there been a wave of delusional people on Hacker News completely neglecting new advancements in technology? The two most common technologies I see having this type of discourse are AI coding and containers.
Either everyone here is a low level quantum database 5D graphics pipeline developer with a language from the future that AI hasn't yet learned, or some people are in denial.
The consequences of incorrect code can be severe outside of front-end web development. For front-end web development, if the code is wrong, you see from your browser that your local web app is broken and try to fix it, or ship it anyway if it's a minor UI bug. For critical backend systems, subtle bugs are often discovered in downstream systems by other teams, and can result in financial loss, legal risk, reputational damage, or even loss of life.
I'm primarily an embedded firmware developer. Gas/electric power products. Ada codebase, so it's off the beaten path but nothing academic by any stretch of the imagination. I have a comprehensive reference manual that describes exactly how the language should be working, and don't need an LLM to regurgitate it to me. I have comprehensive hardware and programming manuals for the MCUs I program that describe exactly how the hardware should be working, and don't need and LLM to regurgitate it to me. I actually really specifically don't want the information transformed, it is engineered to be the way it is, and to change its presentation strips it of a lot of its power.
I deal with way too much torque and way too much electrical energy to trust an LLM. Saving a few minutes here and there isn't worth blowing up expensive prototypes or getting hurt over.
I would be willing to live and let live for the sake of being practical, if the tolerance for (and even active drive towards) low quality slop didn't keep pushing further and further into places it shouldn't. People that accept it in sloppy web dev will accept it in fairly important line of business software. People that accept it in fairly important line of business software will accept it in IT infrastructure. People that accept it in IT infrastructure will accept it in non-trivial security software. People that accept it in non-trivial security software will accept it in what should be a high-integrity system, at which point real engineers or regulatory bodies hopefully step in to stop the bullshit. When asked, everybody will say they draw the line at security, but the drive towards Worse knows no bounds. It's why we see constant rookie mistakes in every IoT device imaginable.
My actual idealistic position, discounting the practicality, is that it shouldn't be tolerated anywhere. We should be trying to minimize the amount of cheap, born-to-die, plasticy shit in society, not maximize it. Most people going on about "muh feature velocity" are reinventing software that has existed for decades. The next shitty UI refresh for Android or Windows, or bad firmware update for whatever device is being screwed up for me, will leave me just as unhappy as the last. The sprint was indeed completed on time, but the product still sucks.
A guided missile should obviously not miss its target. An airliner should obviously never crash. An ERP system should obviously never screw up accounting, inventory, etc, although many people will tolerate that to an unreasonable degree. But my contention is that a phone or desktop's UI should never fail to function as described. A "smart" speaker should never fail to turn on or be controlled. A child's toy should never fail to work in the circumstances they would play with it.
If it's going to constantly fuck up and leave me unhappy and frustrated, why was it made? Why did I buy it? AI could have brought it to market faster, but for what? Once I noticed this, I did just quit buying/dealing with this junk. I'm an ideologue and maybe even a luddite, but I just don't need that bad juju on my soul. I use and write software that's worth caring about.
It’s totally valid to see a new piece of tech, try it, say it’s not for you, and move on. With LLMs it feels forced-fed, and simply saying “eh I’m good, no thanks” isn’t enough. Lots of hype and headlines on how it’s going to take our jobs and replace us, pressure from management to adopt it.
Some new trends make perfect sense to me and I’ll adopt them. I’ve let some pass me by and rarely regretted it. That doesn’t make me a luddite.
I think it’s just backlash against all the AI hype - I get it, im tired of hearing about it too, but - it’s already here to stay, it’s been that way for years now - it’s a normal part of development now for most people, the same as any new tool that becomes the industry darling. Learn to like it or at least learn it, but the reality is here whether you like it or not.
I know my work and it largely isn't shoddy. I have a keen eye for detail and code quality is incredibly important to me. But yeah, I am lazy and I hate wasting my time. AI has been a huge boon in the amount of time it's saved me.
> And the vast majority of people using them are either too stupid or too lazy to actually review their own output.
I don't know if that's true or not. But I'm not stupid or too lazy to review the code, because I review every line and make sure I understand everything. The same way I do with every line of my own code or every line a colleague writes if it's relevant to what I'm working on.
You're in the wrong place if you want to talk about people, particularly fellow developers, in this way. You're just being toxic.
This is a classic case of inflating your own ego and intelligence and just assuming all devs other than you are inferior.
In reality there is a place and time for "lazy and shoddy code." Writing code is always a trade off between taking some amount of tech debt and getting the job done quickly vs writing great code.
I understand not wanting to use LLMs that with no correctness guarantees that randomly hallucinate, but what's wrong with ordinary LSPs and autocompletion? Those seem like perfectly useful tools.
sorry for the snark, but missing the forest for the trees here. unless it's just some philosophical idea, use the tools that save you time. if anything it saves you writing boilerplate or making careless errors.
i don't need to "hand write" every line and character in my code and guess what, it's still easy to understand and maintain because it's what would have written anyway. that or you're just bikeshedding minor syntax.
like if you want to be proud of a "hand built" house with hammer and nails be my guest, but don't conflate the two with always being well built.
Why are you cheapening the product, butchering the process and decimating any hope for further skill development by using these tools?
Instead of python, you should be using assembly or heck, just binary. Instead of relying on an OS abstraction layer made by someone else, you should write everything from scratch on the bare metal. Don't lower yourself by using a text editor, go hex. Then your code will truly be "hand crafted". You'll have even more reason to be proud.
I am unironically with you. I think people should start to learn from computer architecture and assembly and only then, after demonstrating proper skill, graduate to C, and after demonstrating skill there graduate to managed-memory languages.
I was lucky enough to start my programming journey coding in Assembler on the much, much simpler micro computers we had in my youth. I would not even vaguely know where to start with Assembler on a modern machine. We had three registers and a single contiguous block of addressable memory ffs. Likewise, the things I was taught about computer architecture and the fetch-execute cycle back in the 80's are utterly irrelevant now.
I think if you tried to start people off on the kinds of things we started off on in the 80's, you'd never get past the first lesson. It's all so much more complex that any student would (rightly!) give up before getting anywhere.
This comment presupposes that AI is only used to write code that the (presumably junior-level) author doesn’t understand.
I’m a self-respecting software developer with 28 years of experience. I would, with some caveats, venture to say I am an expert in the trade.
AI helps me write good code somewhere between 3x and 10x faster.
This whole-cloth shallow dismissal of everything AI as worthless overhyped slop is just as tired and content-free as breathless claims of the limitless power or universal applicability of AI.
Why use a high level language like python? Why not assembly? Are you really proud of the slow unoptimized byte code that’s executed instead of perfectly crafting the assembly implementation optimizing for the architecture? /s
Seriously comments like yours assume, that all the rest of us who DO make extensive use of these AI tools and have also been around the block for a while, are idiots.
Job market for knowledge jobs isn’t even that good anymore and plenty of people expect it to get worse regardless of their stance on AI. What makes you so sure that LLM users have a bank to laugh all the way to? Already there are many like you, the money you’d make is peanuts
It's funny seeing all of the comments trying to blame the users for this screwup by claiming they're using it wrong. It is reddit though, so I guess I shouldn't be surprised.
what is it about reddit that causes this behavior, when they otherwise are skeptical only of whatever the "official story" is at all costs? it is fascinating behavior.
One reason is lazy one liners are allowed and have high cost/benefit for shitposters and attract many upvotes, so this gets voted highly setting the tone for flamewars in the thread.
It’s miles better on HN. Most bad responses are penalized. The culture is upvoting things that are contributing. I frequently upvote responses that disagree with me. Oftentimes I learn something from it.
Cursor is weird. They have a basically unused GitHub with a thousand unanswered Issues. It's so buggy in ways that VSCode isn't. I hate it. Also I use it everyday and pay for it.
That's when you know you've captured something, when people hate use your product.
Any real alternatives? I've tried continue and was unimpressed with the tab completion and typing experience (felt like laggy typing on a remote server).
VS Code with standard copilot for tab completion and Aider in a terminal window for all the heavier lifts, asking questions, architecting etc. And it’s cheap! I’ve been using it with OpenRouter (lets you easily switch models and providers) and my $10 of credits lasted weeks. Granted, I also use Claude a lot in the browser.
The reason many prefer Cursor over VSCode + GitHub Copilot is because of how much faster Cursor is for tab completion. They use some smaller models that are latency optimized specifically to make the tab completion feel as fast as possible.
Agreed. My laptop has never used swap until I started using cursor… it’s a resource hog, I dislike using it, but it’s still the best AI coding aid and for the work I’m doing right now, the speed boost is more valuable than hand crafted code in enough cases that it’s worth it for me. But I don’t enjoy using the IDE itself, and I used vscode for a few years.
Personally, I will jump ship to Zed as soon as it’s agent mode is good enough (I used Zed as a dumb editor for about a year before I used cursor, and I love it)
I found the Windsurf agent to be relatively less capable, but their inline tool (and the “Tab” they’re promoting so much) has been extremely underwhelming, compared to Cursor.
The only one in this class to be even worse in my experience is Github Copilot.
Cursor + Vim plugin never worked for me, so I switched back to Nvim and never looked back. Nvim already has: avante, codeCompanion, copilot, and many other tools + MCP + aider if you're into that.
Last I heard their team was still 10 people. Best size for doing something revolutionary. Way too few people to triage all that many issues and provide support.
They have enough revenue to hire, they probably are just overwhelmed. They'll figure it out soon I bet.
Any competing product has to absolutely nail tab autocomplete like Cursor has. It's super fast, very smart (even guessing across modules) and very often correct.
so the old adage no such thing as bad PR shows to be incorrect. had they not been in the news, they'd at least have gotten one more monthly sub from you!
This is where Kagi’s subscription policy comes in handy. If you don’t use it for a month, you don’t pay for it that month. There is no need to cancel it and Kagi doesn’t have to pay user acquisition costs.
Slack does this as well. It's a genius idea from a business perspective. Normally IT admins have to go around asking users if they need the service (or more likely you have to request a license for yourself), regularly monitor usage, deactivate stale users etc., all to make sure the company isn't wasting money. Slack comes along and says - don't worry, just onboard every user at the company. If they don't log in and send at least N messages we won't bill them for that month.
They mention an user taking an action will be billed. I guess even sending a message or reacting with an emoji would count as taking an action ? Even logging in ?
Cursor is trapped in a cat and mouse game against "hacks" where users create new accounts and get unlimited use. The repo was even trending on Github (https://github.com/yeongpin/cursor-free-vip).
Sadly, Cursor will always be hampered by maintaining it's own VSCode fork. Others in this niche are expanding rapidly and I, myself, have started transitioning to using Roo and Cline.
literally any service with a free trial--i.e. literally any service--has this "problem". it's an integral part of the equation in setting up free trials in the first place, and by no means a "trap". you're always going to have a % of users who do this, the business model relies on the users who forget and let the subscription cross over to the next month or simply feel its worth paying
This is true but Cursor’s problems are a bit worse than a normal paywalled service.
Cursor allows users to get free credits without a credit card and this forced them to change their VSCode fork on how it handles identification so they can stop users from spawning new accounts.
Another is that normally, companies have a cost for each free user. For Cursor, this cost is so sporadic since it doesn’t charge per million context, they use credits. Free users get 50 credits but 1 credit could be 200k+ context each so it could be $40-50 per free user per month. And these users get 50 credits every month.
Lastly, the cursor vip free repo has trended on GitHub many times and users who do pay might stop and use this repo instead.
The Cursor vip free creator is well within his rights to do what they want and get “free” access. This unfortunately hurts paying customers since Cursor has to stop these “hacks.”
This is why Cursor should just move to a VSCode extension. I’ve used Augment and other VSCode extensions and the feature set is close to Cursor so it’s possible for them just to be an extension. The other would be to remove free accounts but allow users to bring their own keys. To use Composer/Agent, you can’t bring your own keys.
This will allow Cursor to stop maintaining a VSCode fork, helps them stop caring if users create new accounts (since all users are paying) and lets users bring their own keys if they don’t want to pay. Hell, if they charge a lifetime fee to bring our own keys for Agent, that would bring in revenue too. But as I see now, Roo and Cline’s agent features are catching up and Cursor won’t have a moat soon.
> Cursor is trapped in a cat and mouse game against "hacks" where users create new accounts and get unlimited use
Actually, you don't even have to make a new account. You can delete your account and make it again reusing the same email.
I did this on accident once because I left the service and decided to come back, and was surprised to get a free tier again. I sent them an email letting them know that was a bug, but they never responded.
I paid for a month of access just to be cautious, even though I wasn't using it much. I don't understand why they don't fix this.
There will be a fork-a-month for these products until they have the same lockin as a textbox that you talk at, "make million dollar viral facebook marketplace post"
What is the evidence that "dozens of users publicly canceled their subscriptions"?
A total of 4 users claimed that they did or would cancel their subscriptions in the comments, and 3/4 of them hedged by saying that they would cancel if this problem were real or happened to them.
It looks like only 1 person claimed to have cancelled already.
Submitted title was "Cursor IDE support hallucinates lockout policy causes mass user cancellations" - I've de-massed it now.
Since the HN title rule is "Please use the original title, unless it is misleading or linkbait" and the OP title is arguably misleading, I kept the submitter's title. But if there's a more accurate or neutral way to say what happened, we can change it again.
Currently, there are 59 comments after the post was deleted and the thread was locked.
It is worth mentioning that the comments that remain are valuable, as they highlight the captured market size and express concern about the impending deterioration of the situation.
One workaround for using RAG, as mentioned in a podcast I listened to, involves employing a second LLM agent to assess the work of the first LLM. This agent evaluates the response or hallucination by requiring the first LLM to cite sources and subsequently locate those sources.
trying to "fix hallucinations" is like trying to fix humans being wrong. it's never going to happen. we can maybe iterate towards an asymptote, but we're never going to "fix hallucinations"
No: an LLM that doesn't confabulate will certainly get things wrong in some of the same ways that honest humans do - being misinformed, confusing similar things, "brain" damage from bad programming or hardware errors. But LLM confabulations like the one we're discussing only occur in humans when they're being sociopathically dishonest. A lawyer who makes up a court case is not a "human being wrong," it's a human lying, intentionally trying to deceive. When an LLM does it, it's because it is not capable of understanding that court cases are real events that actually happened.
Cursor's AI agent simply autocompleted a bunch of words that looked like a standard TOU agreement, presumably based on the thousands of such agreements in its training data. It is not actually capable of recognizing that it made a mistake, though I'm sure if you pointed it out directly it would say "you're right, I made a mistake." If a human did this, making up TOU explanations without bothering to check the actual agreement, the explanation would be that they were unbelievably cynical and lazy.
It is very depressing that ChatGPT has been out for nearly three years and we're still having this discussion.
Some time ago, in an electronics online shop I asked about warranty terms from a chatbot and got favorable answers. It didn't take two hours before a human contacted me via email to correct a misunderstanding.
I have an uneasy feeling logging into a Text Editor vscode and seeing a Microsoft correlated account, work or personal, in the lower left corner. I understand that settings sync or whatever but it’d be preferred to keep to a simple config json or xml (pretty sure most settings are in json).
I have no problem, however, pasting an encryption public key into my Sublime Text editor. I’m not completely turned off by ability fir telemetry, tracking, or analytics. But having a login for a Text Editor is totally unappealing to me with all the overhead.
It’s a bummer that similar to browsers and chrome, the text editor with an active package marketplace necessitates some tech major underwriting the development with “open source” code but a closed kernel.
Long live Sublime text (i’m aware there are more pure text editors but do use mice)
Consider the source. I assume companies will learn the hard way not to fake human interaction and be forced to mention when responses are auto-generated (a better word than "AI" really). All that effort for so many to build rapport and trust and all pissed away by trying to save a buck and having an ambitious bot trying to go solo.
Wow. This one will go down in the history books as yet another example of AI hype outpacing AI capability.
FTFY
Also see every single genAI PR release showing obvious uncanny valley image (hands with more than expected number of fingers). See Apple's propaganda videos vs actual abilities. There are plenty of other (all???) PR examples where the product does not do what is advertised on the tin.
I've had a very good experience with Cursor on small Typescript projects.
It started hallucinating a lot as my typescript project got bigger.
I found it pretty useless in languages like Go and C++.
I ended up canceling Cursor this month. It was messing up working code, suggesting random changes, and ultimately increasing my cognitive load instead of reducing it.
The current AIs are such people pleasers, that I really take the time how to form the prompt, so that it can't just say, "yes, it is like that" in a polite way.
LLMs for customer support is for ghetto companies that want to cheap out on quality. That's why you'll see Comcast and such use it, for example, but not your broker or anywhere where the stakes on the company's reputation are non-zero.
I don't understand the negativity. I use Cursor and love it.
Are there real challenges with forking VS Code? Yep. Are there glitches with LLMs? Sure. Are there other AI-powered coding alternatives that can do some of the same things? You betcha.
But net-net, Cursor's an amazing power tool that strongly extends what we can accomplish in any hour, day, or week.
I was playing around with ChatGPT the other day and encountered a similar issue. Once I hit the rate limit, the replies telling me that seemed to be AI generated. When I waited the requisite time, the next use attempt repeated the last message about being past the limit and needing to wait 13 hours again.
It seemed to be reading from the conversation to determine this. Oops! Replaying an earlier message worked fine.
Literally the only safe way to use an LLM in a business context is as an input to a trusted human expert, and the jury is still out for even that case.
Letting an AI pose as customer support is just begging for trouble, and Cursor had their wish appropriately granted.
god I hate redditors. what is it about that website that makes every user so so incredibly desperate to jump on literally any bandwagon they can lay their woolly arses on?
This drama is a very good thing because 1) now companies might reconsider replacing customer support with AI and 2) the initial victim was an AI company.
It could be better though.. I wish this happened to a company providing "AI support solutions"..
It's an AI company that is presumably drowning in money. What humans do work there have probably already had a good laugh and forgotten about this incident.
Are you using a locally ran LLM that is equally capable as Claude 4.7? Kind of seems like the answer has to be "not as capable and also the hardware investment was insane"
Incidentally, this is a great way to find out you're never going to get real support when you need it, just AI responses designed to make you get tired of trying to talk to a real person they need to pay.
AI's killer app is "marketing". Bots are incredibly useful for selling items, services, and politics, and AI makes the bots indistinguishable from real people to most people most of the time. It's highly effective so I don't see that market shrinking any time soon.
There’s a bubble, in the sense of the dot-com bubble. There’s a lot of money being thrown at the AI equivalents of Pets.com, and that will pop eventually. But the internet persisted and transformed every aspect of the society even after the bubble. I think AI will be similar.
1) there is no AI bubble, it has revolutionized how we communicate and learn
2) you don't need to buy an expensive GPU for local LLM, any contemporary laptop with enough RAM is sufficient to run an uncensored Gemma-3 fast
Railroads revolutionized transport and yet railway mania was undeniably a bubble. Something can be both very useful and yet also overvalued and overhyped leading to significant malinvestment (sometimes, everyone wins to the detriment of the investors, sometimes just everyone loses out because a huge amount of effort was spent on not useful stuff, usually somewhere in between).
I'm an AI fan, but there's clearly a desperate attempt by just about every tech company to integrate AI at the cost of genuinely productive developments in the space, in a manner that one might describe as a "bubble." Microsoft's gotta buy enough GPUs to handle Copilot in Windows Notepad, after all...
Calling it desperate is a subjective assessment. Yes, some strategies are more haphazard than others, but ignoring generative AI currently is the same as ignoring the internet in 1999 or mobile in 2010 (which facebook famously regretted and paid $4+1B to buy instagram and whatsapp in order to catch up)
I'm shocked by this perspective, and I'm deep into the LLM game (shipped 7 figure products using LLMs). I don't feel like anything has been revolutionized around communication - I can spot AI generated emails pretty easily (just send the prompt, people). On the learning front I do find LLMs to be more capable search engines for many tasks, so they're helpful absolutely.
I love this moment so much, I want to marry it and raise a family of little moments with it. A company that claims AI will solve everything cannot even get their own support right. chefkiss
Cursor is bevoming useless now. They are too money greedy to be trusted. Last time i installed on another machine tgey told me to subscrribe to use. It is gross.
1. Whenever AI is used closed loop, with human feedback and a human checking the output for quality/hallucinations and then passing it along, it's often fine.
2. Whenever it is used totally on its own, with no humans in the loop, it's awful and shit like this happens.
Yet, every AI company seems to want to pretend we're ready for #2, they market their products as #2, they convince their C-suite customers that their companies should buy #2, and it's total bullshit--we're so far from that. AI tools can barely augment a human in the driver's seat. It's not even close to being ready to operate on its own.
Good. I hope other companies stupid enough to subject their users to unleashed AI for support instead of real humans reap the consequences of their actions. I'll be eating popcorn and mocking them from my little corner of the internet.
Awesome. It warms my heart that both parties who are supporting AI, users and company, felt its negative effects. We can only hope this continues to happen.
The original bug that prevented login was no doubt generated by AI. They probably ran it through an AI code review tool like the one posted here recently.
The world is drowning in bullshit and delusion. Programming was one of the few remaining places where you had to be precise, where it was harder to fool yourself. Where you had to understand it to program it. That's being taken away and it looks like a lot of people are embracing what is coming. It's hardly surprising - we just love our delusions too much.
at no point did they say anything that even comes close to saying “no bugs in human written code.”
if you’re willing to come down off your defensive AI position, because your response is a common one from people who are bought into the tech, i’ll try explain what they were saying (if not, stop reading now, save yourself some time).
maybe you’ll learn something, who knows :shrug:
> Programming was one of the few remaining places where you had to be precise, where it was harder to fool yourself. Where you had to understand it to program it.
they are talking about the approach, motivations and attitudes involved in “the craft”.
we strive for perfection, knowing we will never reach it. we, as programmers/hackers/engineers must see past our own bullshit/delusions to find our way to the fabled “solution”.
they are lamenting how those attitudes have shifted towards “fuck it, that’ll do, who cares if the code reads good, LLM made it work”.
where in the “vibe coding” feedback loop is there a place for me, a human being, to realise i have completely misunderstood a concept for the last five years and suddenly realise “oh shit, THATS HOW THAT WORKS!? HOW HAVE I NOT REALISED THAT FOR FIVE YEARS.” ?
where in “just ask chatgpt for a summary about a topic” is my journey where i learn about a documentation rendering library that i never even knew existed until i actually started reading the docs site for a library?
maybe we were thinking about transferring our docs off confluence onto a public site to document our API? asking chatGpt removes that opportunity for accidental learning and growth.
in essence, they’re lamenting the sacrifice people seem to be willing to make for convenience, at the price of continually growing and learning as a human being.
at least that’s my take on it. probably wrong — but if i am at least i get to learn something new and grow as a person and see past my own bullshit and delusions!
Obligatory reference to the Streisand Effect: that trying to hide something (e.g. a Reddit post) often has the unintended consequence of drawing more attention to it.
Oooh I wanted to try this for a while so here goes…
This doesn’t seem like anything new. Ill-informed support staff has always existed, and could also give bad information to users. AI is not the problem. And it hasn’t created any problems that weren’t already there before AI.
Usually by the time I get to a post on HN criticizing AI, someone has already posted this exact type of rebuttal to any criticism…
I haven't seen people comment on just the wow factor here. Apparently Cursor produced a full integrated AI app and it orchestrated a self-destruct process in fashion emulating the way some human-managed companies have recently self-destructed. AI fails are easy and some require a lot of work, apparently.
Looking forward to apps trained on these Reddit threads.
Who will be the first AI-first company to close the loop with autonomous coding bots, and create self-fulfilling prophecies where a customer support bot hallucinates a policy like OP, it gets stored in the Q&A/documentation retrieval database, and the autonomous bots implement and roll out the policy and lock out all the users (and maybe the original human developers as well)?
It's what we all wanted. Replacing your human support team to be run exclusively by AI LLM bots whilst they hallucinate to their users. All unchecked.
Now this bug has now turned into a multi-million dollar mistake and costed Cursor to lose millions of dollars overnight.
What if this was a critical control system in a hospital or energy company and their AI support team (with zero humans) hallucinated a wrong meter reading and overcharged their customers? Or the AI support team hallucinated the wrong medication to a patient?
Are we sure that the damage is in “millions”? Could it not be just the same old “vocal and loud minority”, who will wake up tomorrow and re-activate their subscription before the lunch break once Cursor team comes and writes a heartfelt apology post (well generated by AI with a system prompt of “you are a world class PR campaign manager …”)?
You're right, but this is more general - it has nothing to do with AGI and everything to do with poor management. It reminds me very much of the Chernobyl disaster, and the myriad examples in Taleb's "The Black Swan".
But look, AI did take a job today. The job of cursor product developers is gone now. ;) But seriously, every time I hear media telling me AI is going to replace us, they conveniently forget episodes like this. AI may take the jobs one day, but it doesn't seem like that day is any time soon when AI adopters keep getting burned.
From cursor developer: "Hey! We have no such policy. You're of course free to use Cursor on multiple machines.
Unfortunately, this is an incorrect response from a front-line AI support bot. We did roll out a change to improve the security of sessions, and we're investigating to see if it caused any problems with session invalidation. We also do provide a UI for seeing active sessions at cursor.com/settings.
and the best thing about it is that the base model is going to be trained on reddit posts so expect SupportBot3000 to be even more confident about this fact in the future!
I mean to be fair, I like that they're putting their money where their mouth is so to speak - if you want to sell a product based on the idea that AI can handle complex tasks, you should probably have AI doing what should be simple, frontline support.
> you should probably have AI doing what should be simple, frontline support.
AI companies are going to prove (to the market or to the actual people using their products) that a bunch of "simple" problems aren't at all simple and have been undervalue for a long time.
I don't agree with that at all. Hallucination is a very well known issue. Sure leverage AI to improve their productivity.. but not even having a human look over the responses shows they don't care about their customers
The number of times real human powered support caused me massive headache and sometimes financial damage and the number of times my lawyer fixed those because me trying to explain why they were wrong… I am not surprised that AI will do the same as the creation is the image of the creator and all that.
If you had a human support person feeding the support question into the AI to get a hint, do you think that support person is going to know that the AI response is made up and not actually a correct answer? If they knew the correct answer, they wouldn't have needed to ask the AI.
> if you want to sell a product based on the idea that AI can handle complex tasks, you should probably have AI doing what should be simple, frontline support.
That would only be true if you were correct that your AI can handle complex tasks. If you want to sell dowsing rods, you probably don't want to structure your own company to rely on the rods.
Which is basically an article to use an extension in a way that’s basically forbidden use.
If that was not bad enough the editor also told you to install certain extensions if certain file extensions were used that were also against the tos of the extension.
And basically cursor can just be using the vsix marketplace from eclipse, which does not contain restricted extensions.
What they do is at least shady.
And yes I’m not a fan of the fact that Microsoft does this, even worse they closed the source (or some parts of it) of some extensions as well, which is also a bad move (but their right)
The extensions themselves have licenses that prohibit their use with anything other than VSCode.
(You should keep this in mind next time someone tells you that VSCode is "open source", by the way. The core IDE is, sure, but if you need to do e.g. Python or C++, the official Microsoft extensions involved all have these kinds of clauses in them.)
I don't use VSCode (or Cursor in this case (which I do think was malicious in the way it blindly hallucinated a policy for a paying customer)); I use vim or notepad++ depending on my mood.
I just don't have a problem with people "violating" Terms of Service or End User License Agreements and am not really convinced there's a legal argument there either.
For distribution licenses, I would assume they have. Can't put GPL software in your closed source code, can't just download Photoshop and copy it and give it out, etc. And that makes sense and you have some reasonable path to damage/penalties (GPL → your software is now open source, Photoshop → fines or whatever)
But if you download some free piece of software and use it with some other piece of free piece software even though they say "please don't" in the EULA, what could the criminal or civil penalties possibly be?
I don't know what the hypothetical penalty would be for mere use contrary to EULA, though. It would be breach of contract, and presumably the court would determine actual damages, but I don't know what cost basis there would be if the software in question was distributed freely. However, fine or no fine, I would expect the court to order the defendant to cease using software in violation of EULA, and at that point further use would be contempt of court, no?
Interesting, but isn't it just because Teams is bundled and integrated with MS Office? I couldn't find any specific stats on revenue or how many businesses actually choose to pay for Teams specifically.
Microsoft pushed Teams onto my personal Windows PC in a recent update, as a startup item. And, as far as I could tell, automatically logged me in to my Microsoft account on it.
I'd argue that humans also more easily learn from huge mistakes. Typically, we need only one training sample to avoid a whole class of errors in the future (also because we are being held accountable).
How is this not an example of humans being held accountable? What would be the difference here if a help center article contained incorrect information? Would you go after the technical writer instead of the founders or Cursor employees responding on Reddit?
As annoying as it is when the human support tech is wrong about something, I'm not hoping they'll lose their job as a result. I want them to have better training/docs so it doesn't happen again in the future, just like I'm sure they'll do with this AI bot.
That only works well if someone is in an appropriate job though. Keeping someone in a position they are unqualified for and majorly screwing up at isn't doing anyone any favors.
> I'm not hoping they'll lose their job as a result
I have empathy for humans. It's not yet a thought crime to suggest that the existence of an LLM should be ended. The analogy would make me afraid of the future if I think about it too much.
I probably should have added sarcasm tags to my post. My very firm opinion is that AI should only make suggestions to humans and not decisions for humans.
Fingers crossed the users will go left are all the noisy ones. I still enjoy using Cursor but their forum is filled with too many posts about $20 being too expensive or they need to raise caps or Cursor is the worse tool in the world.
Yes, I’m complaining about complainers. It’s the circle of life on Hacker News, someone always has to play the food chain’s top predator: the mildly annoyed power user.
Which of course they are going to try to brush it all away. Better than admitting that this problem very much still exists and isn’t going away anytime soon.
The section about hallucinations is deeply relevant.
Namely, Claude sometimes provides a plausible but incorrect chain-of-thought reasoning when its “true” computational path isn’t available. The model genuinely believes it’s giving a correct reasoning chain, but the interpretability microscope reveals it is constructing symbolic arguments backward from a conclusion.
https://en.wikipedia.org/wiki/On_Bullshit
This empirically confirms the “theory of bullshit” as a category distinct from lying. It suggests that “truth” emerges secondarily to symbolic coherence and plausibility.
This means knowledge itself is fundamentally symbolic-social, not merely correspondence to external fact.
Knowledge emerges from symbolic coherence, linguistic agreement, and social plausibility rather than purely from logical coherence or factual correctness.
To me it feels like people that benefit from or at least enjoy that sort of assistance and I solve vastly different problems and code very differently.
I’ve done exhausting code reviews on juniors’ and middles’ PRs but what I’ve been feeling lately is that I’m reviewing changes introduced by a very naive poster. It doesn’t even type-check. Regardless of whether it’s Claude 3.7, o1, o3-mini, or a few models from Hugging Face.
I don’t understand how people find that useful. Yesterday I literally wasted half an hour for a test suite setup a colleague of mine introduced to the codebase that wasn’t good, and I tried delegating that fix to several of the Copilot models. All of them missed the point, some even introduced security vulnerabilities in the process invalidating JWT validation, I tried “vide coding” it till it works, until I gave up in frustration and just used an ordinary search engine, which led me to the docs, in which I immediately found the right knob. I reverted all that crap and did the simple and correct thing. So my conclusion was simple: vibe coding and LLMs made the codebase unnecessarily more complicated and wasted my time. How on earth do people code whole apps with that?
Yes they didn't push it as hard as, say, copilot. I still think they got in way too deep way too fast.
Yes they knew Apple maps was bad and not up to standard yet, but they didn't really have any other choice.
The models and devices just aren't quite there yet.
Once Google gets its shit together and starts deploying (cloud--based) AI features to Android devices en masse, Apple is going to have a really big problem on their hands.
Most users say that they want privacy, but if privacy comes in the way of features or UX, they choose the latter. Successful privacy-respecting companies (Apple, Signal) usually understand this, it's why they're successful, but I think Apple definitely chose the wrong tradeoff here.
I don’t necessarily agree with the post you’re responding to, but what I will give Apple credit for is making their AI offering unobtrusive.
I tried it, found it unwanted and promptly shut it off. I have not had to think about it again.
Contrast that with Microsoft Windows, or Google - both shoehorning their AI offering into as many facets of their products as possible, not only forcing their use, but in most cases actively degrading the functionality of the product in favor of this required AI functionality.
Henry Ford was 23 years “slow” relative to Karl Benz.
Apple fumbled a bit with Siri, and I'm guessing they're not too keen to keep chasing everyone else, since outside of limited applications it turns out half baked at best.
Sadly, unless something shinier comes along soon, we're going to have to accept that everything everywhere else is just going to be awful. Hallucinations in your doctor's notes, legal rulings, in your coffee and laundry and everything else that hasn't yet been IoT-ified.
I installed a logitech mouse driver (sigh) the other day, and in addition to being obtrusive and horrible to use, it jams an LLM into the UI, for some reason.
AI has reached crapware status in record time.
I was in the VC space for a while previously, most pitch decks claimed to be using AI: But doing even the briefest of DD - it was generally BS. Now it's real.
With respect to everything being awful: One might say that's always been the case. However, now there's a chance (and requirement) to build in place safeguards/checks/evals and massively improve both speed and quality of services through AI.
Don't judge for the problems: Look at the exponential curve, think about how to solve the problems. Otherwise, you will get left behind.
Every week for the last few months, I get a recruiter for a healthcare startup note taking app with AI. It's just a rehash of all the existing products out there, but "with AI". It's the last place I want an overworked non-technical user relying on the computer to do the right thing, yet I've had at least four companies reach out with exactly that product. A few have been similar. All of them have been "with AI".
It's great that it is getting better, but at the end of the day, there's only so much it can be relied upon for, and I can't wait for something else to take away the spotlight.
Yes. Finally! Now it's real BS. I wouldn't touch it with 8 meter pole.
[0] https://www.bbc.com/news/articles/cq5ggew08eyo
I'm certain they'll get it right soon enough though. People were writing off Google in terms of AI until this year.. and oh how attitudes have changed.
Just give Google a year or two.
Google has a pretty amazing history of both messing up products generally and especially "ai like" things, including search.
(Yes I used to defend Google until a few years ago.)
that's the problem noo?? big company is sucks at that, you cant do that in certain company because sometimes its just not possible
There is no world in which a compiler or tooling will save you from the absolute mayhem it can do. I’ve had it routinely try to re-implement third party libraries, modify code unrelated to what it was asked, quietly override functions etc.
It’s like a developer who is on LSD.
But in reality hallucinations either make people using AI lose a lot of their time trying to stuck the LLMs from dead ends or render those tools unusable.
Don't get me wrong, I use AI every day, but it's mostly as a localized code complete or to help me debug tricky issues. Meaning I've written and understand the code myself, and the AI is there to augment my abilities. AI works great if it's used as a deductive tool.
Where it runs into issues is when it's used inductively, to create things that aren't there. When it does this, I feel the hallucinations can be off the charts -- inventing APIs, function names, entire libraries, and even entire programming languages on occasion. The AI is more than happy to deliver any kind of information you want, no matter how wrong it is.
AI is not a tool, it's a tiny Kafkaesque bureaucracy inside of your codebase. Does it work today? Yes! Why does it work? Who can say! Will it work tomorrow? Fingers crossed!
I use AI for rather complex tasks. It's impressive. It can make a bunch of non-trivial changes to several files, and have the code compile without warnings. But I need to iterate a few times so that the code looks like what I want.
That being said, I also lose time pretty regularly. There's a learning curve, and the tool would be much more useful if it was faster. It takes a few minutes to make changes, and there may be several iterations.
It sounds like the guys in this article should not have trusted AI to go fully open loop on their customer support system. That should be well understood by all "customers" of AI. You can't trust it to do anything correctly without human feedback/review and human quality control.
2) No matter what the learning curve, you're using a statistical tool that outputs in probabilities. If that's fine for your workflow/company, go for it. It's just not what a lot of developers are okay with.
Of course it's a spectrum with the AI deniers in one corner and the vibe coders in the other. I personally won't be relying 100% on a tool and letting my own critical thinking atrophy, which seems to be happening, considering recent studies posted here.
Not op but yes. It sometimes takes a lot of time but I read everything. It still faster than nothing. Also, I ask very precise changes to the AI so it doesn’t generate huge diffs anyway.
Also for new code, TDD works wonders with AI : let it write the unit tests (you still have to be mindful of what you want to implement) and ask it to implement the code that run the tests. Since you talk the probabilistic output, the tool is incredibly good at iterating over things (running and checking tests) and also, unit tests are, in themselves, a pretty perfect prompt.
Opposite experience for me. It reliably fails at more involved tasks so that I don't even try anymore. Smaller tasks that are around a hundred lines maybe take me longer to review that I can just do it myself, even though it's mundane and boring.
The only time I found it useful is if I'm unfamiliar with a language or framework, where I'd have to spend a lot of time looking up how to do stuff, understand class structures etc. Then I just ask the AI and have to slowly step through everything anyways, but at least there's all the classes and methods that are relevant to my goal and I get to learn along the way.
2) I find the tool analogy helpful but it has limits. Yes, it’s a stochastic tool, but in that sense it’s more like another mind, not a tool. And this mind is neither junior nor senior, but rather a savant.
This is just an incredible statement. I can't think of another development tool we'd say this about. I'm not saying you're wrong, or that it's wrong to have tools we can't just, just... wow... what a sea change.
Imagine if your calculator app randomly and non-deterministically performed arithmetic incorrectly, and you similarly couldn't get correctness expectations from the developer.
Imagine if any of your communication tools randomly and non-deterministically translated your messages into gibberish...
I think we'd all throw away such tools, but we are expected to accept it if it's an "AI tool?"
They keep telling you that any employee who highlights problems with the tools are just trying to save their job.
Your investors tell you that the toolmakers are already saving money for your competitors.
Now, do you want that second house and white lotus vacation or not?
Making good tools is difficult. Bending perception (“is reality”) is easier and enterprise sales, just like good propaganda, work. The gold rush will leave a lot of bodies behind but the shovelmakers will make a killing.
If you think of AI like a programmer, no we shouldn't throw away such tools because we accept them as imperfect and we still need to review.
This is a common argument but I don't think it holds up. A human learns. If one of my teammates or I make a mistake, when we realize it we learn not to make that mistake in the future. These AI tools don't do that. You could use a model for a year, and it'll be just as unreliable as it is today. The fact that they can't learn makes them a nonstarter compared to humans.
Unreliable tools have a good deal of utility. That's an example of them helping reduce the problem space, but they also can be useful in situations where having a 95% confidence guess now matters more that a 99.99% confidence one in ten minutes- firing mortars in active combat, say.
There's situations where validation is easier than computation; canonically this is factoring, but even division is much simpler than multiplication. It could very easily save you time to multiply all of the calculator's output by the dividend while performing both a multiplication and a division for the 5% that are wrong.
edit: I submit this comment and click to go the front page and right at the top is Unsure Calculator (no relevance). Sorry, I had to mention this
This is generally true when you can quantify the unreliability. E.g. random prime number tests with a specific error rate can be combined so that the error rates multiply and become negligible.
I'm not aware that we can quantify the uncertainty coming out of LLM tools reliably.
The problem is that you don't know which 5% are wrong. The AI is confidently wrong all the time. So the only way to be sure is to double check everything, and at some point its easier to just do it the right way.
Sure, some things don't need to be perfect. But how much do you really want to risk? This company thought a little bit of potential misinformation was acceptable, and so it caused a completely self inflicted PR scandal, pissed off their customer base, and lost them a lot of confidence and revenue. Was that 5% error worth it?
Stories like this are going to keep coming the more we rely on AI to do things humans should be doing.
Someday you'll be affected by the fallout of some system failing because you happen to wind up in the 5% failure gap that some manager thought was acceptable (if that manager even ran a calculation and didn't just blindly trust whatever some other AI system told them) I just hope it's something as trivial as an IDE and not something in your car, your bank, or your hospital. But certainly LLMs will be irresponsibly shoved into all three within the next few years, if it's not there already.
No, you still have to do 100% of the work.
Of course there is a review system for a reason, but we frequently use "untrusted" tools in development.
That one guy in a github issue that said "this worked for me"
If there is such a tool, programmers will be on path of immediate reskilling or lose their jobs very quickly.
Because no other dev tool actually generates unique code like AI does. So you treat it like the other components of your team that generates code, the other developers. Do you trust other developers to write good code without mistakes without getting it reviewed by others. Of course not.
And then you say "by the time the rest of the team reviews it. Most code review is uneventful."
So you trust your team to develop without the need for code review but yet, your team does code review.
So what is the purpose of these code reviews? Is it the case that you actually don't think they are necessary, but perhaps management insists on them? You actually answer this question yourself:
> Most code review is uneventful.
Keyword here is "most" as opposed to "all" So based your team's applied practices and your own words, code review is for the purpose of catching mistakes and other needed corrections.
But it seems to me if you trust your team not to make mistakes, code review is superfluous.
As an aside, it seems your team culture doesn't make room for juniors because if your team had juniors I think it would be even more foolish to trust them not to make mistakes. Maybe a junior free culture works for your company, but that's not the case for every company.
My main point is code review is not superfluous no matter the skill level; junior, senior, or AI simply because everyone and every AI makes mistakes. So I don't trust those three classes of code emitters to not ever make mistakes or bad choices (i.e. be perfect) and therefore I think code review is useful.
Have some honesty and humility and you'll amazed at what's possible.
The purpose of the review is to find and fix occasional small details before it goes to physical testing. It does not involve constant babysitting of the developer. It's a little silly to bring up honesty when you spent that entire comment dancing around the reality that AI makes an inordinately large number of mistakes. I will pick the domain expert who refuses to touch AI over a generic programmer with access to it ten times out of ten.
The entire team as it is now (me included) were juniors. It's a traditional engineering environment in a location where people don't aggressively move between jobs at the drop of a hat. You don't need to constantly train younger developers when you can retain people.
> I "trust other developers to write good code without mistakes without getting it reviewed by others". Of course I can trust them to do the right thing even when nobody's looking, and review it anyway in the off-chance they overlooked something.
You're saying yes, I trust other developers to not make mistakes, but I'll check anyways in case they do. If you really trusted them not to make mistakes, you wouldn't need to check. They (eventually) will. How can I assert that? Because everyone makes mistakes.
It's absurd to expect anyone to not make mistakes. Engineers build whole processes to account for the fact that people, even very smart people make mistakes.
And it's not even just about mistakes. Often times, other developers have more context, insight or are just plain better and can offer suggestions to improve the code during review. So that's about teamwork and working together to make the code better.
I fully admit AI makes mistakes, sometimes a lot of them. So it needs code review . And on the other hand, sometimes AI can really be good at enhancing productivity especially in areas of repetitive drudgery so the developer can focus on higher level tasks that require more creativity and wisdom like architectural decisions.
> I will pick the domain expert who refuses to touch AI over a generic programmer with access to it ten times out of ten.
I would too, but I won't trust them not to make mistakes or occasional bad decisions because again, everybody does.
> You don't need to constantly train younger developers when you can retain people.
But you do need to train them initially. Or do you just trust them to write good code without mistakes too?
Literally yes. Test coverage and QA to catch bugs sure but needing everything manually reviewed by someone else sounds like working in a sweatshop full of intern-level code bootcamp graduates, or if you prefer an absolute dumpster fire of incompetence.
Building on AI seems more like building on a foundation of sand, or building in a swamp. You can probably put something together, but it's going to continually sink into the bog. Better to build on a solid foundation, so you don't have to continually stop the thing from sinking, so you can build taller.
Am I arguing in favor of egalitarian commit food fights with no adults in the room? Absolutely not. But demanding literally every change go through a formal review process before getting committed, like any other coding dogma, has a tendency to generate at least as much bullshit as it catches, just a different flavor.
Then it's not a useful tool, and I will decline to waste time on it.
It does poorly without heavy instruction, though, especially with anything more than toy projects.
Still a valuable tool, but far from the dreamy autonomous geniuses that they often get described as.
This is not an inherent flaw of LLMs, rather it is a flaw of a particular implementation-if you use guided sampling, so during sampling you only consider tokens allowed by the programming language grammar at that position, it becomes impossible for the LLM to generate ungrammatical output
> When it does this, I feel the hallucinations can be off the charts -- inventing APIs, function names, entire libraries,
They can use guided sampling for this too - if you know the set of function names which exist in the codebase and its dependencies, you can reject tokens that correspond to non-existent function names during sampling
Another approach, instead of or as well as guided sampling, is to use an agent with function calling - so the LLM can try compiling the modified code itself, and then attempt to recover from any errors which occur.
Your tone is rather hyperbolic here, making it sound like an extra brace resulted in a disaster. It didn't. It was easy to detect and easy to fix. Not a big deal.
When a tool starts confidently inserting random wrong code into my 100% correct code, there's not much more I need to see to know it's not a tool for me. That's less like a tool and more like a vandal. That's not something I need in my toolbox, and I'm certainly not going to replace my other tools with it.
Apologies - something very clearly went wrong here. We’ve already begun investigating, and some very early results:
* Any AI responses used for email support are now clearly labeled as such. We use AI-assisted responses as the first filter for email support.
* We’ve made sure this user is completely refunded - least we can do for the trouble.
For context, this user’s complaint was the result of a race condition that appears on very slow internet connections. The race leads to a bunch of unneeded sessions being created which crowds out the real sessions. We’ve rolled out a fix.
Appreciate all the feedback. Will help improve the experience for future users.
Don't use AI. Actually care. Like, take a step back, and realise you should give a shit about support for a paid product.
Don't get me wrong: AI is a very effective tool, *for doing things you don't care about*. I had to do a random docker compose change the the other day. It's not production code, it will be very obvious whether or not AI output works, and I very rarely touch docker and don't care to become a super expert in it. So I prompted the change, and it was good enough and so I ran with it.
You using AI for support tells me that you don't care about support. Which tells me whether or not I should be your customer.
I agree with you, they should care.
(Today I learned)
This person is not the only one to experiencing this bug. As this thread has pointed out.
HN goes a step further. It has a function that allows moderators to kill or boost a post by subtracting or adding a large amount to the post's score. HN is primarily a place for Y Combinator to hype their latest venture, and a "safe" place for other startups and tech companies.
The team I work with it took nearly 3 months to get basic questions answered correctly when it came to a sales contract. They never gave our Sec team acceptable answers around privacy and security.
Reddit is free to play for marketing firms. Perhaps they could add extra statistics, analytics, promotions for these commercial users.
Edit: he did refund 22 mins after seeing this
The best case scenario is that you lied about having people answer support. LLMs pretending to be people (you named it Sam!) and not labeled as such is clearly intended to be deceptive. Then you tried to control the narrative on reddit. So forgive me if I hit that big red DOUBT button.
Even in your post you call it "AI-assisted responses" which is as weaselly as it gets. Was it a chatbot response or was a human involved?
But 'a chatbot messed up' doesn't explain how users got locked out in the first place. EDIT: I see your comment about the race condition now. Plausible but questionable.
So the other possible scenario is that you tried to hose your paying customers then when you saw the blowback blamed it on a bot.
'We missed the mark' is such a trope non-apology. Write a better one.
I had originally ended this post with "get real" but your company's entire goal is to replace the real with the simulated so I guess "you get what you had coming". Maybe let your chatbots write more crap code that your fake software engineers push to paying customers that then get ignored and/or lied to when they ask your chatbots for help. Or just lie to everyone when you see blowback. Whatever. Not my problem yet because I can write code well enough that I'm embarrassed for my entire industry whenever I see the output from tools like yours.
This whole "AI" psyop is morally bankrupt and the world would be better off without it.
Also, illegal in the EU.
Slightly related to this; I just wanted to ask whether all Cursor email inboxes are gated by AI agents? I've tried to contact Cursor via email a few times in the past, but haven't even received an AI response :)
Cheers!
Are you truely that cheap? Is this why it took you guys 3 months to get a basic contract back to us?
AI fixes most of that... Most of the time? Clearly not, but hey.
Don’t let the dickish replies get to you.
I was offered in writing a refund by the team who cold reached out to me to ask me why I cancelled my sub one week after start. Then they ignored my 3+ emails in response asking them to refund, and other means of trying to communicate with them. Offering me a refund as a bait to gain me back, then when I accept it they ghost me. Wow. Very low.
The product is not terrible but the team responses are. And this, if you see how they handled it, is also a very poor response. First thing you notice if you open the link is that the Cursor team removed the reddit post! As if we were not going to see it or something? Who do they think they are? Censoring bad comments which are 100% legit.
I am giving it a go to competitors just out of sheer frustration with how they handle customers, and I do recommend everybody to explore other products before you settle on Cursor. I don't intend to ever re-subscribe and have recommended friends to do the same, most of which agree with my experience.
You just don't know how to prompt it correctly.
Note that I have absolutely no knowledge or reason to believe this other than general distrust of companies.
Yeah, who puts an AI in charge of support emails with no human checks and no mention that it's an AI generated reply in the response email?
The bot is now called "UltralyticsAssistant" and discloses that it's automated, which is welcome. The bad advice is all still there though.
(I don't know if they're really _famous_ for this, but among friends and colleagues I have talked to multiple people who independently found and were frustrated by the useless github issues.)
VibeSupport bots, how well did that work out for Canada Air?
https://thehill.com/business/4476307-air-canada-must-pay-ref...
More evidence: all of the ignorant uses of "hallucinate" here, when what's happening is FABRICATION.
I believe it’s pronounced VibeOops.
"Caution: Any of this could be wrong."
Then again paying users might wonder "what exactly am I paying for then?"
It also seems like a weird thing to lie about, since it's just another very public example of AI fucking up something royally, coming from a company whose whole business model is selling AI.
Which is crazy. Support is part of marketing so it should get the same kind of consideration.
Why do people think Amazon is hard to beat? Price? nope. Product range? nope. Delivery time? In part. The fact if you have a problem with your product they'll handle it? Yes. After getting burned multiple times by other retailers you're gonna pay the Amazon tax so you don't have to ask 10 times for a refund or be redirected to the supplier own support or some third party repair shop.
Everyone knows it. But people are still stuck on the "support is a cost center" way of life so they keep on getting beat by the big bad Amazon.
The thing is, what the AI hallucinated (if it was an AI-hallucinating), was the kind of sleezy thing companies do do. However, the thing with sleezy license changes is they only make money if the company publicizes them. Of course, that doesn't mean a company actually thinks that far ahead (X many managers really think "attack users ... profit!"). Riddles in enigmas...
I'll just keep chugging along, with debian, python and vim, as I always have. No LLM, no LSP, heck not even autocompletion. But damn proud of every hand crafted, easy to maintain and fully understood line of code I'll write.
Now, I don't trust the output - I review everything, and it often goes wrong. You have to know how to use it. But I would never go back. Often it comes up with more elegant solutions than I would have. And when you're working with a new platform, or some unfamiliar library that it already knows, it's an absolute godsend.
I'm also damn proud of my own hand-crafted code, but to avoid LLMs out of principal? That's just luddite.
20+ years of experience across game dev, mobile and web apps, in case you feel it relevant.
Getting to sit down and write the code is the most enjoyable part of the job, why would I deprive myself of that? By the time the problem has been defined well enough to explain it to an LLM sitting down and writing the code is typically very simple.
Code review is difficult to get right, especially if the goal is judging correctness. Maybe this is a personal failing, but I find being actively engaged to be a critical part of the process; the more time I spend with the code I'm maintaining (and usually on call for!) the better understanding I have. Tedium can sometimes be a great signal for an abstraction!
I can imagine that LLM is really helpful in some cases for some people. But so far, I couldn’t find a single example when I and simple copy-pasting wouldn’t have been faster. Not even when I tried it, not when others showed me how to use it.
Don't get me started on testcase generation.
What I've found frustrating about the narrative around these tools; I've watched them from afar with intrigue but ultimately found that method of working just isn't for me. Over the years I've trialed more tools than I can remember and adopted the ones I found useful, while casting aside ones that aren't a great fit. Sometimes I find myself wandering back to them once they're fully baked. Maybe that will be the case here, but is it not valid to say "eh...this isn't it for me"? Am I kidding myself?
And yet the time it takes me to use the LLM and correct its output is usually faster than not using it at all.
Over time I've developed a good sense for what tasks it succeeds at (or is only trivially wrong) and what tasks it's just not up for.
The folly of single ended metrics.
> but to avoid LLMs out of principal? That's just luddite.
Do you double check that the LLM hasn't magically recreated someone else's copyrighted code? That's just irresponsible in certain contexts.
> in case you feel it relevant.
Of course it's relevant. If a 19 year old with 1 year of driving experience tries to sell me a car using their personal anecdote as a metric I'd be suspicious. If their only salient point is that "it gets me to where I'm going faster!" I'd be doubly suspicious.
I frankly do not care, and I expect LLMs to become such ubiquitous table-stakes that I don't think anyone will really care in the long run.
In python I was scanning 1000’s of files each for thousands of keywords. A naive implementation took around 10 seconds, obviously the largest share of execution time after running instrumentation. A quick ChatGPT led me to Aho-Corasick and String searching algorithms, which I had never used before. Plug in a library and bam, 30x speed up for that part of the code.
I could have asked my knowledgeable friends and coworkers, but not at 11PM on a Saturday.
I could have searched the web and probably found it out.
But the LLM basically auto completed the web, which I appreciate.
Once I had to look up a research paper to implement a computational geometry algorithm because I couldn't find it any of the typical Web sources. There were also no library to use with a license for our commercial use.
I'm not against use of "AI". But this increasing refusal of those who aspire to work in specialist domains like software development to systematically learn things is not great. That's just compounding on an already diminished capacity to process information skillfully.
Many developers use libraries effectively without knowing every time consideration of O(n) comes into play.
Competently implemented, in the right context, LLMs can be an effective form of abstraction.
Get friends with weirder daily schedules. :-)
I've had a long-term code project that I've really struggled with, for various reasons. Instead of using my normal approach, which would be to lay out what I think the code should do, and how it should work, I just explained the problem and let the LLM worry about the code.
It got really far. I'm still impressed. Claude worked great, but ran out of free tokens or whatever, and refused to continue (fine, it was the freebie version and you get what you pay for). I picked it up again in Cursor and it got further. One of my conditions for this experiment was to never look at the code, just the output, and only talk to the LLM about what I wanted, not about how I wanted it done. This seemed to work better.
I'm hitting different problems, now, for sure. Getting it to test everything was tricky, and I'm still not convinced it's not just fixing the test instead of the code every time there's a test failure. Peeking at the code, there are several remnants of previous architectural models littering the codebase. Whole directories of unused, uncalled, code that got left behind. I would not ship this as it is.
But... it works, kinda. It's fast, I got a working demo of something 80% near what I wanted in 1/10 of the time it would have taken me to make that manually. And just focusing on the result meant that I didn't go down all the rabbit holes of how to structure the code or which paradigm to use.
I'm hooked now. I want to get better at using this tool, and see the failures as my failures in prompting rather than the LLM's failure to do what I want.
I still don't know how much work would be involved in turning the code into something I could actually ship. Maybe there's a second phase which looks more like conventional development cleaning it all up. I don't know yet. I'll keep experimenting :)
Sir, you have just passed vibe coding exam. Certified Vibe Coder printout is in the making but AI has difficulty finding a printer. /s
Because I can ship 2x to 5x more code with nearly the same quality.
My employer isn't paying me to be a craftsman. They're paying me to ship things that make them money.
Either way, LLMs are actually high up the quality spectrum as they generate a very consistent style of code for everyone. Which gives it uniformity, that is good when other developers have to read and troubleshoot code.
This definition limits the number of problems you can solve this way. It basically means buildup of the technical debt - good enough for throwaway code, unacceptable for long term strategy (growth killer for scale-ups).
>Either way, LLMs are actually high up the quality spectrum
This is not what I saw, it’s certainly not great. But that may depend on stack.
The ones with self-respect aren't the ones using it
I think the creators of Redis and Django are very capable and self-respecting software developers.
Either everyone here is a low level quantum database 5D graphics pipeline developer with a language from the future that AI hasn't yet learned, or some people are in denial.
I deal with way too much torque and way too much electrical energy to trust an LLM. Saving a few minutes here and there isn't worth blowing up expensive prototypes or getting hurt over.
My actual idealistic position, discounting the practicality, is that it shouldn't be tolerated anywhere. We should be trying to minimize the amount of cheap, born-to-die, plasticy shit in society, not maximize it. Most people going on about "muh feature velocity" are reinventing software that has existed for decades. The next shitty UI refresh for Android or Windows, or bad firmware update for whatever device is being screwed up for me, will leave me just as unhappy as the last. The sprint was indeed completed on time, but the product still sucks.
A guided missile should obviously not miss its target. An airliner should obviously never crash. An ERP system should obviously never screw up accounting, inventory, etc, although many people will tolerate that to an unreasonable degree. But my contention is that a phone or desktop's UI should never fail to function as described. A "smart" speaker should never fail to turn on or be controlled. A child's toy should never fail to work in the circumstances they would play with it.
If it's going to constantly fuck up and leave me unhappy and frustrated, why was it made? Why did I buy it? AI could have brought it to market faster, but for what? Once I noticed this, I did just quit buying/dealing with this junk. I'm an ideologue and maybe even a luddite, but I just don't need that bad juju on my soul. I use and write software that's worth caring about.
Some new trends make perfect sense to me and I’ll adopt them. I’ve let some pass me by and rarely regretted it. That doesn’t make me a luddite.
> And the vast majority of people using them are either too stupid or too lazy to actually review their own output.
I don't know if that's true or not. But I'm not stupid or too lazy to review the code, because I review every line and make sure I understand everything. The same way I do with every line of my own code or every line a colleague writes if it's relevant to what I'm working on.
You're in the wrong place if you want to talk about people, particularly fellow developers, in this way. You're just being toxic.
In reality there is a place and time for "lazy and shoddy code." Writing code is always a trade off between taking some amount of tech debt and getting the job done quickly vs writing great code.
Anyways, Cursor generates all my code now.
i don't need to "hand write" every line and character in my code and guess what, it's still easy to understand and maintain because it's what would have written anyway. that or you're just bikeshedding minor syntax.
like if you want to be proud of a "hand built" house with hammer and nails be my guest, but don't conflate the two with always being well built.
Why are you cheapening the product, butchering the process and decimating any hope for further skill development by using these tools?
Instead of python, you should be using assembly or heck, just binary. Instead of relying on an OS abstraction layer made by someone else, you should write everything from scratch on the bare metal. Don't lower yourself by using a text editor, go hex. Then your code will truly be "hand crafted". You'll have even more reason to be proud.
I think if you tried to start people off on the kinds of things we started off on in the 80's, you'd never get past the first lesson. It's all so much more complex that any student would (rightly!) give up before getting anywhere.
Don’t get too hung up on what works for other people. That’s not a good look.
I’m a self-respecting software developer with 28 years of experience. I would, with some caveats, venture to say I am an expert in the trade.
AI helps me write good code somewhere between 3x and 10x faster.
This whole-cloth shallow dismissal of everything AI as worthless overhyped slop is just as tired and content-free as breathless claims of the limitless power or universal applicability of AI.
Seriously comments like yours assume, that all the rest of us who DO make extensive use of these AI tools and have also been around the block for a while, are idiots.
It’s miles better on HN. Most bad responses are penalized. The culture is upvoting things that are contributing. I frequently upvote responses that disagree with me. Oftentimes I learn something from it.
This is true for the entirety of Reddit, and the majority is deranged.
That's when you know you've captured something, when people hate use your product.
Any real alternatives? I've tried continue and was unimpressed with the tab completion and typing experience (felt like laggy typing on a remote server).
Personally, I will jump ship to Zed as soon as it’s agent mode is good enough (I used Zed as a dumb editor for about a year before I used cursor, and I love it)
The only one in this class to be even worse in my experience is Github Copilot.
https://github.com/RooVetGit/Roo-Code
I use Zed with `3.7 sonnet`.
They have enough revenue to hire, they probably are just overwhelmed. They'll figure it out soon I bet.
Sadly, Cursor will always be hampered by maintaining it's own VSCode fork. Others in this niche are expanding rapidly and I, myself, have started transitioning to using Roo and Cline.
Cursor allows users to get free credits without a credit card and this forced them to change their VSCode fork on how it handles identification so they can stop users from spawning new accounts.
Another is that normally, companies have a cost for each free user. For Cursor, this cost is so sporadic since it doesn’t charge per million context, they use credits. Free users get 50 credits but 1 credit could be 200k+ context each so it could be $40-50 per free user per month. And these users get 50 credits every month.
Lastly, the cursor vip free repo has trended on GitHub many times and users who do pay might stop and use this repo instead.
The Cursor vip free creator is well within his rights to do what they want and get “free” access. This unfortunately hurts paying customers since Cursor has to stop these “hacks.”
This is why Cursor should just move to a VSCode extension. I’ve used Augment and other VSCode extensions and the feature set is close to Cursor so it’s possible for them just to be an extension. The other would be to remove free accounts but allow users to bring their own keys. To use Composer/Agent, you can’t bring your own keys.
This will allow Cursor to stop maintaining a VSCode fork, helps them stop caring if users create new accounts (since all users are paying) and lets users bring their own keys if they don’t want to pay. Hell, if they charge a lifetime fee to bring our own keys for Agent, that would bring in revenue too. But as I see now, Roo and Cline’s agent features are catching up and Cursor won’t have a moat soon.
Actually, you don't even have to make a new account. You can delete your account and make it again reusing the same email.
I did this on accident once because I left the service and decided to come back, and was surprised to get a free tier again. I sent them an email letting them know that was a bug, but they never responded.
I paid for a month of access just to be cautious, even though I wasn't using it much. I don't understand why they don't fix this.
It makes number go up and to the right
In a corporate environment, compliance needs are far more important than some trivial cost.
A total of 4 users claimed that they did or would cancel their subscriptions in the comments, and 3/4 of them hedged by saying that they would cancel if this problem were real or happened to them. It looks like only 1 person claimed to have cancelled already.
Is there some other discussion you're looking at?
Since the HN title rule is "Please use the original title, unless it is misleading or linkbait" and the OP title is arguably misleading, I kept the submitter's title. But if there's a more accurate or neutral way to say what happened, we can change it again.
Yet if you went by the HN comments, you'd think it were the biggest item on primetime news.
People are really champing at the bit.
It is worth mentioning that the comments that remain are valuable, as they highlight the captured market size and express concern about the impending deterioration of the situation.
> Apologies about the confusion here.
If this was a sincere apology, they'd stop trying to make a chat bot do support.
Nope, there's no reliable solution for them, as of yet.
There's hope that hallucinations will be solved by someone, somehow, soon... but hope is not a strategy.
There's also hype about non-stop progress in AI. Hype is more a strategy... but it can only work for so long.
If no solution materializes soon, many early-adopter LLM projects/trials will be cancelled. Sigh.
Cursor's AI agent simply autocompleted a bunch of words that looked like a standard TOU agreement, presumably based on the thousands of such agreements in its training data. It is not actually capable of recognizing that it made a mistake, though I'm sure if you pointed it out directly it would say "you're right, I made a mistake." If a human did this, making up TOU explanations without bothering to check the actual agreement, the explanation would be that they were unbelievably cynical and lazy.
It is very depressing that ChatGPT has been out for nearly three years and we're still having this discussion.
I have no problem, however, pasting an encryption public key into my Sublime Text editor. I’m not completely turned off by ability fir telemetry, tracking, or analytics. But having a login for a Text Editor is totally unappealing to me with all the overhead.
It’s a bummer that similar to browsers and chrome, the text editor with an active package marketplace necessitates some tech major underwriting the development with “open source” code but a closed kernel.
Long live Sublime text (i’m aware there are more pure text editors but do use mice)
- github integration (e.g. git auth, sync text editor settings in private gist)
- a trusted third party server for negotiating p2p sessions with someone else (for pair programming, debugging over a call, etc...)
But anyone who wants to remove the microsoft/github account features from their editor entirely can just use vscodium instead.
FTFY
Also see every single genAI PR release showing obvious uncanny valley image (hands with more than expected number of fingers). See Apple's propaganda videos vs actual abilities. There are plenty of other (all???) PR examples where the product does not do what is advertised on the tin.
No AI, less crappy frameworks, fewer dark patterns, etc.
It started hallucinating a lot as my typescript project got bigger.
I found it pretty useless in languages like Go and C++.
I ended up canceling Cursor this month. It was messing up working code, suggesting random changes, and ultimately increasing my cognitive load instead of reducing it.
It seems you're not aware of the issue which plagued tens of Cursor releases, where the software would auto-delete itself on updates.
It was pretty hilarious, to be honest. Your workflow would consist of always installing the editor before use.
Are there real challenges with forking VS Code? Yep. Are there glitches with LLMs? Sure. Are there other AI-powered coding alternatives that can do some of the same things? You betcha.
But net-net, Cursor's an amazing power tool that strongly extends what we can accomplish in any hour, day, or week.
It seemed to be reading from the conversation to determine this. Oops! Replaying an earlier message worked fine.
Letting an AI pose as customer support is just begging for trouble, and Cursor had their wish appropriately granted.
It could be better though.. I wish this happened to a company providing "AI support solutions"..
surely it wouldn't backfire, right???
ok aside from the joke from this case alone, I think we can all agree that AI not replacing human soon
Maybe this time is different. Maybe not.
2. Whenever it is used totally on its own, with no humans in the loop, it's awful and shit like this happens.
Yet, every AI company seems to want to pretend we're ready for #2, they market their products as #2, they convince their C-suite customers that their companies should buy #2, and it's total bullshit--we're so far from that. AI tools can barely augment a human in the driver's seat. It's not even close to being ready to operate on its own.
Except they did make a mistake: trusting their Simulated Intelligence (I’m done calling it “AI”.) with their customers’ trust.
The world is drowning in bullshit and delusion. Programming was one of the few remaining places where you had to be precise, where it was harder to fool yourself. Where you had to understand it to program it. That's being taken away and it looks like a lot of people are embracing what is coming. It's hardly surprising - we just love our delusions too much.
if you’re willing to come down off your defensive AI position, because your response is a common one from people who are bought into the tech, i’ll try explain what they were saying (if not, stop reading now, save yourself some time).
maybe you’ll learn something, who knows :shrug:
> Programming was one of the few remaining places where you had to be precise, where it was harder to fool yourself. Where you had to understand it to program it.
they are talking about the approach, motivations and attitudes involved in “the craft”.
we strive for perfection, knowing we will never reach it. we, as programmers/hackers/engineers must see past our own bullshit/delusions to find our way to the fabled “solution”.
they are lamenting how those attitudes have shifted towards “fuck it, that’ll do, who cares if the code reads good, LLM made it work”.
where in the “vibe coding” feedback loop is there a place for me, a human being, to realise i have completely misunderstood a concept for the last five years and suddenly realise “oh shit, THATS HOW THAT WORKS!? HOW HAVE I NOT REALISED THAT FOR FIVE YEARS.” ?
where in “just ask chatgpt for a summary about a topic” is my journey where i learn about a documentation rendering library that i never even knew existed until i actually started reading the docs site for a library?
maybe we were thinking about transferring our docs off confluence onto a public site to document our API? asking chatGpt removes that opportunity for accidental learning and growth.
in essence, they’re lamenting the sacrifice people seem to be willing to make for convenience, at the price of continually growing and learning as a human being.
at least that’s my take on it. probably wrong — but if i am at least i get to learn something new and grow as a person and see past my own bullshit and delusions!
https://en.wikipedia.org/wiki/Streisand_effect
There's windsurf, cline, zed, copilot got a huge update too, is cursor still leading the space?
This doesn’t seem like anything new. Ill-informed support staff has always existed, and could also give bad information to users. AI is not the problem. And it hasn’t created any problems that weren’t already there before AI.
Usually by the time I get to a post on HN criticizing AI, someone has already posted this exact type of rebuttal to any criticism…
Makes you think of that one meme.
Looking forward to apps trained on these Reddit threads.
Use AI to augment but don't really replace it as a 100% system if you can't predict and own up the failure rate.
My advice would be to use more configurable tools with less interest on selling fake perfection. Aider works.
It's what we all wanted. Replacing your human support team to be run exclusively by AI LLM bots whilst they hallucinate to their users. All unchecked.
Now this bug has now turned into a multi-million dollar mistake and costed Cursor to lose millions of dollars overnight.
What if this was a critical control system in a hospital or energy company and their AI support team (with zero humans) hallucinated a wrong meter reading and overcharged their customers? Or the AI support team hallucinated the wrong medication to a patient?
Is this the AGI future we all want?
Unfortunately, this is an incorrect response from a front-line AI support bot. We did roll out a change to improve the security of sessions, and we're investigating to see if it caused any problems with session invalidation. We also do provide a UI for seeing active sessions at cursor.com/settings.
Apologies about the confusion here."
AI companies are going to prove (to the market or to the actual people using their products) that a bunch of "simple" problems aren't at all simple and have been undervalue for a long time.
Such as support.
That would only be true if you were correct that your AI can handle complex tasks. If you want to sell dowsing rods, you probably don't want to structure your own company to rely on the rods.
https://www.cursor.com/how-to-install-extension
Which is basically an article to use an extension in a way that’s basically forbidden use.
If that was not bad enough the editor also told you to install certain extensions if certain file extensions were used that were also against the tos of the extension.
And basically cursor can just be using the vsix marketplace from eclipse, which does not contain restricted extensions.
What they do is at least shady.
And yes I’m not a fan of the fact that Microsoft does this, even worse they closed the source (or some parts of it) of some extensions as well, which is also a bad move (but their right)
(You should keep this in mind next time someone tells you that VSCode is "open source", by the way. The core IDE is, sure, but if you need to do e.g. Python or C++, the official Microsoft extensions involved all have these kinds of clauses in them.)
I just don't have a problem with people "violating" Terms of Service or End User License Agreements and am not really convinced there's a legal argument there either.
For distribution licenses, I would assume they have. Can't put GPL software in your closed source code, can't just download Photoshop and copy it and give it out, etc. And that makes sense and you have some reasonable path to damage/penalties (GPL → your software is now open source, Photoshop → fines or whatever)
But if you download some free piece of software and use it with some other piece of free piece software even though they say "please don't" in the EULA, what could the criminal or civil penalties possibly be?
I don't know what the hypothetical penalty would be for mere use contrary to EULA, though. It would be breach of contract, and presumably the court would determine actual damages, but I don't know what cost basis there would be if the software in question was distributed freely. However, fine or no fine, I would expect the court to order the defendant to cease using software in violation of EULA, and at that point further use would be contempt of court, no?
Cursor is at a worse position and at greater risk of ending up like Slack very quickly and Microsoft will do the exact same thing they did to Slack.
This time by extinguishing (EEE) them by racing prices of VSCode + Copilot close to zero, until it is free.
The best thing Cursor should do is for OpenAI to buy them at a $10B valuation.
I'd be very skeptical of their MAU claims.
I have empathy for humans. It's not yet a thought crime to suggest that the existence of an LLM should be ended. The analogy would make me afraid of the future if I think about it too much.