I... don't know how I feel about this. I mean, the concept, sure, good idea. Is it reliable? Perhaps. But... I get a "buy merch" popup the moment I open the page, and there seems to be a pro option for what seems to be a very much LLM-aided app? ...IDK, rubs off a bit wrong IMO. Just my personal opinion.
We are going to have the organic label soon on apps with this mindset..AI assisted development is evolving by the day and at the rate most apps will be built that way..
If the developer can't be bothered to not make it look vibecoded I assume there's a minimal amount of thought and effort put into it. And in that case I'm not going to bother using it.
Maybe this developer didn't bother I didn't even look frankly.
My point was that you can have very well designed and thoughted apps with LLMs. In fact, I'd argue you can go way further and deeper than any manual coded application paired with AI assistance.
But maybe this instance is not the case. But even if it turns out to be not the case, if the app does work and serves the utility you are after, then why not using it? People use many things in everyday life without thinking about what thought was put into it..in fact, most people don't think at all about what it took to build anything, and how it was built..
Do you really know or care if the shoes you wearing was handmade or built by a factory in china using robots? you probably don't know and don't care.
> My point was that you can have very well designed and thoughted apps with LLMs.
If that were the case here, then a user of the app probably would not come away with the impression that it was a "very much LLM-aided app."
You're talking about how LLMs can be used in such a way that they help production of an app but remain invisible to the end user. But the beginning of this thread was someone using the app and coming away with the impression that it was heavily LLM influenced, which suggests that it was not done thoughtfully and did not remain invisible to the user.
I get it, and frankly I don't know about this app.
But the parent comment positioned itself against all "LLM-aided apps", the author didn't say a slop or poorly coded. And even if it looks like LLM coded, we still can't tell about the level of thought that went into it. Developers had many poorly coded UIs and yet the apps functioned well and served it's purpose.
So the entire chain of reasoning is a form stereotype forming. Looking at this specific app, I can see that the author thought about the marketing/business and functionality, more than other alternatives, maybe because they had more mental bandwdith to do so.
To be fair, I personally only said that I'd like a label that indicates whether an app has "had thought put into it". That could involve e.g. another human reviewing the code and architecture and giving it a stamp.
I hand written each one of them, I think they are full of typos and grammatical errors as well since English is not my first language.
I assume you are saying that because I apologized. I apologize for sounding like AI, I assure you it was hand written and I put thoughts into those comments.
I don’t see how this comment is “against all LLM-aided apps.”
> But... I get a "buy merch" popup the moment I open the page, and there seems to be a pro option for what seems to be a very much LLM-aided app? ...IDK, rubs off a bit wrong IMO. Just my personal opinion.”
Again, if the user of the app comes away with the impression that the app “seems to be a very much LLM-aided” then the LLM usage has affected the end product.
I think people are judging mainly the UI of the app. This is closed code app so they don't know, but they are judging by patterns in the UI and/or wording of documentation. And I think it is mainly tech or artists or whoever else is impacted by the AI usage.
That is not say there are really mindlessly produced apps out there. But the comment made it sound that anything LLM-aided is negative.
But what does this has to do with LLM-assisted coding? I wouldn't trust this app even it was manually coded.
In fact, given the recent reports, in terms of security, one is better off consulting LLMs on the exposure surfaces. Something a solo developer can now do cheaply but in the past required a lot of specialized knowledge.
> But what does this has to do with LLM-assisted coding?
What does "LLM-assisted coding" have to do with the question I answered? Here's that question again:
> if the app does work and serves the utility you are after, then why not using it?
Note that this question does not contain any reference to "LLMs". Indeed, an LLM-generated product can contain adware, spyware, or trojan horses just like a human-written one.
For additional context, this app in question seems to actually contain adware and spyware, and has a closed source codebase, so might contain a trojan horse as well.
Those seem like pretty good reasons to not run this totally-vibe-coded software, so I'm sorry, but I don't understand your argument that LLMs make this okay or not okay or whatever.
My comment was about not equating every LLM assisted app as poorly thought of. Your comment is about this specific app, and I personally would not install this app regardless of how it was coded.
But again in terms of security since you brought it, I'd argue that having LLM assisted coded app might be secure than manually coded application given the recent advancements. Again, this is a generic statement, not specific to this app.
The original discussion was not about every LLM assisted app, it was specifically about vibe-coded apps, particularly ones with so little human input that experienced coders can tell they were vibe-coded.
Your later comments generalizing to "every LLM assisted app" seem like a bit of insecurity you might be feeling for using LLM assistance. No need to feel that way. LLM assistance is okay. We're discussing something different from mere LLM assistance.
That is the original comment: " for what seems to be a very much LLM-aided app? "
The author mentioned LLM-aided app, not even AI assisted; nor vibe coded hence my reaction. I don't have insecurity from using LLM nor from coding manually. But I noticed a trend in people thinking and biases and it threw me off hence the comment.
My original comment was heavily downvoted and furthermore some of my comments were accused of being written by AI meanwhile I'm writing it by hand. Again, people are taking tribes/camps, if you speaking of insecurities, it actually the opposite clearly, many programmers are worried of being displaced and automated. And I do understand the concern, I'm just stating the fact that one can have very deeply thoughted LLM-aided app.
It did make it easier to generate thoughtless apps, I agree.
But this process is new, and not every AI assisted app is thoughtless. In fact, I'd argue that if a person put more thoughts and amplify their abilities with AI assisted coding, then the total sum of mental labour exceed that of any manually coded app by huge margins.
I'm not sure about this particular app, haven't looked into it. But I was merely commenting about the mindset of the parent comment.
If one person knows how to code, they don't need an LLM. I've tried using all of them on my job, and I will give them a second chance time by time (to see if the claims made by these big tech hold up) and ALL of the time I've didn't noticed any speedup compared to coding manually.
If you are an experienced programmer (not someone that did an online course and calls like this) with YEARS of experience in coding in multiple languages really, the only real limitation is the speed at which you can type on the keyboard, that is usually way faster than what the even good LLM is able to emit tokens.
This not considering token expense: if we see at the point of view of the company, it's way cheaper an experienced programmer, that in my country costs you 4000 a month tax included, than the tokens that an LLM would use to do the same job.
On top of that there is the environmental impact of all of this, and why they are almost making me not use my car because it's euro 5 and now there is euro 6, these AI companies are emitting tons of CO2 to do stuff that human beings can do with 1/1000 the energy that these AI companies use.
"if one person knows how to code, they don't need an LLM. I've tried using all of them on my job, and I will give them a second chance time by time (to see if the claims made by these big tech hold up) and ALL of the time I've didn't noticed any speedup compared to coding manually."
I mean, have you truly tried? do others thing this is true? I'm truly astonished if that is majority experience, I think we need to get surveys on this.
The entire draw of coding with LLMs is that you can generate programs without knowing how to code. That's literally what "vibe coding" is. Why would it surprise you that people who do know how to code, and don't hate the process the way a lot of AI enthusiasts seem to, might not find them useful?
You'll find plenty of people on HN who have this experience,but they just tend to be ridiculed or told they just haven't used AI correctly or enough. Admitting you don't use LLMs for everything, or god forbid at all, is like telling Christians you don't want to be saved.
Because they can easily get outcompeted by those who care about software/output and use the latest tools? might as well code in assembly and brag about it.
I'm not here condemning people who use or not use. We have a new tech, for anyone who have been in the field long enough, it is nothing short of magic was has been accomplished in the last few years.
I could turn around 10K lines of code per day (I know it's not the best measurement but it does capture size to some extent) when I was not able to pull more than a 100 LOC. Let us say that most of it work, and there are few rough edges, a senior programmer can follow the shape of the code, and understand where to tweak and where the LLM get stuck. So this is basically like having racing engine compared to running.
Companies get outcompeted, individual coders don't. And not everyone wants to be the Thomas Kinkade of coding, just cranking product out as fast as possible. And products that "use the latest tools" don't always "outcompete" those that don't. Most of what gets vibe coded - as demonstrated by what gets posted to Show HN - is qualitatively worse than the equivalent human-written software. It certainly isn't better, and there's no indication that velocity to market leads to market dominance.
To metaquote a recent thread - where is the vibe-coded replacement for Photoshop that outcompetes it? or Word? Why aren't vibe-coded games dominating Steam?
The speed alone doesn't translate to productivity and productivity alone doesn't translate to quality.
Imo it depends on the type of change required. If it involves churning out 1000 line of new boilerplate, an LLM will win every time. If it means tracing a subtle edge-case bug and carefully planning a deep one-line change in a 100k codebase that leaves all other behaviour unaffected, I would still expect an experienced dev who's deeply familiar with the code to do it at least as fast and more correctly first time.
It sounds more like you are equating a totally-vibe coded product with no verified code, with a product thoughtfully written by a human with LLM assistance.
I like the idea of this type of merch that normalizes anti big tech sentiment. But I personally would also opt for something more dramatic, possibly involving a guillotine...on the t shirt I mean
Social media has tuned people to be unable to think in non-hyperbolic ways. To these constantly wound-up people, every problem, no matter how big or small, can only be solved by brutal violence and savagery.
The longer people see issues that are not being solved via peaceful means, the more accepting they become of non-peaceful means.
I'm not advocating for those means, but I understand how people are feeling more helpless than ever to just "let the system work it out" when the system appears to be bought and paid for by a small concentration of corporations and people with an increasingly disproportionate share of wealth.
> To these constantly wound-up people, every problem, no matter how big or small, can only be solved by brutal violence and savagery.
well, the idea of democracy was to offer a viable path for people's discontent to be heard and remediated.
Unfortunately in most Western countries, dysfunctionality and corporatism have taken it over, and that is why you're seeing authoritarians on the rise across the world. They paint democracy and "legacy" figures as the scapegoat and themselves (plus handing them allllllll the power) as the only solution - just look at Milei and the chainsaw.
I wouldn't see myself as constantly wound up, but I get your angle. I also was at most half serious, alluding to a time not long ago in a country not far away, where this instrument had its moment.
And in retrospect, the French revolution is not considered a dark episode of modern history, on the contrary.
The following is 100% vibe coded by me, (except for the algorithm that actually makes the music recommendations, which is about 5 lines of code I wrote 30 years ago.)
I created it in two months using Claude Code. It would have taken me three years without it. I would have never done it.
What about this free service is a win for big tech?:
The thing is, it doesn't even have to be PG/HN doing this. Case in point, look at reddit being astroturfed by every company.
And also, look how good Graphene OS posts do, and now compare that to Stop Killing Games. These both have a similar overlap (unless you work at Google, both are interests/hobbies), yet GOS almost always hits front page (and rightly so), yet SKG failed to, despite being a Stallman-ian topic.
True, but there's that 10% posts where there's (eg) a good cancer breakthrough, and it's two guys going back and forth on the methodology adding a lot of comments. Not always, but not never either.
I really wish there was a way to "mark" such posts as "high quality, ignore comment to vote ratio" just like how you can vouch a comment. Enough such marks and it gets auto turned off the flamewar parameter.
Back-and-forth conversation isn't high quality by definition, it's just the baseline for not being bad enough to downvote. I think Hacker News wants the baseline to be some degree of exceptional.
Reading yet another thread where commenters go back and forth on, eg how Elon Musk sucks/is awesome doesn't make for interesting reading because it's the same conversation that's been had before, and neither side is getting convinced by the other side's argument.
Noob comes from newb, which comes from newbie. Some argue it comes from "nubile" but I don't personally buy that etymology. I distinctly remember "newb" followed by "noob" but that was eons ago.
I mean, both can be true at the same time. For a slightly controversial example, look at Kanye. If you knew him before he got off his rocker (especially the 2000s), he was a very different person than who he has become today.
1. He should probably go back on his bipolar medication, even if he feels less creative on it.
2. Paying for the surgery his mother died during really messed up his head. Partly because he probably blames himself, partly because she was one of the few stabilizing influences in his life. He really started going downhill after.
He might have always been impulsive-grandstanding-interrupt-Taylor-Swift Kanye, but it's conceivable he wouldn't have morphed into slavery-is-a-state-of-mind-"i-love-hitler"-antisemitic Kanye.
Sigh. "I miss the old Kanye." Time to give 808s and Heartbreak a re-listen.
Depends on your phone, but for some qualcomm devices there was a vulnerability some time ago, iirc the OnePlus 15 has a dedicated thread on XDA. I also found this on XDA while trying to find the other exploit:
Side note, it's quite ironic, google being so heavily anti-root is forcing people to opt for root access via more hidden ways, making it easier to avoid root being detected.
Now that you mention it, it does make some sense that with the rising prices (especially for ram) (and maybe with LLMs?) it’s cheaper to use basic hardware and put “more effort” on efficient code (an LLM’s effort, of course).
reply