You think it'll rapidly get smarter, but it just recreates things from all the terrible code it was fed.
Code and how it is written also rapidly changes these days and LLMs have some trouble drawing lines between versions of things and the changes within them.
Sure, they can compile and test things now, which might make the code work and able to run. The quality of it will be hard to increase without manually controlling and limiting the type of code it 'learns' from.
> Is it maintainable? Well it's AI that's going to maintain it.
That's what's currently not possible, it might work in a small webapp or similar.
But in a large system, it absolutely falls apart when having to maintain it.
Sure, it can fix a bug, but it doesn't understand the side effects it creates with the fix, yet.
Maybe in the future that will also be possible. I do agree with you about business/management not caring about long term impacts if short term gains are possible.
I also think this is why AI works okay-ish on tiny new greenfield webapps and absolutely doesn't on large legacy software.
You can't accurately plan every little detail in an existing codebase, because you'll only find out about all the edge cases and side effects when trying to work in it.
So, sure, you can plan what your feature is supposed to do, but your plan of how to do that will change the minute you start working in the codebase.
For MS, it's currently eroding through every single one of their products.
Azure, Office, Visual Studio, VS Code, Windows are all shipping faster than ever, but so much stuff is unfinished, buggy, incompatible to existing things, etc.
If you want a chef to explain and show his recipe and then you find out he asked chat gpt or got it from a recipe website, would you still be interested in hearing about the 'journey' he took?
If that chef then also ordered the food from a catering service by forwarding that recipe to them, does it make sense to interview him about his craft and listen to him?
It's the same with creative things. Sure, you can ask a musician or a painter about their inspiration/process etc., because they went through the process and made something. You could learn something from them, other than 'here's how i asked someone else to make this'.
So, going to an art forum and displaying your AI generated art there, ready to answer questions about the process is pretty much completely pointless and also cheapens every actual artist that came there to display and talk about something they actually made.
It also has a psychological effect on the people that use it. I know some people that get immense feelings of accomplishment out of using AI to generate art and music. They feel like they made something and are proud about it. For a lot of people like that, any incentive to learn about things is gone, because they get the exact same feeling by using AI to make it.
On the other hand: If you are an artist making things and there are a million people generating things every 5 minutes and showing it around everywhere, it dilutes recognition of what you made. You show it to your friends and they're like: Oh yeah, i made like 5 of those yesterday.
Yes I agree with what you are saying. And this is why I am saying that if this place is a place for chef's then I agree that even a minor use of AI can be disheartening.
Once again if HN is solely for the 1st sector of chefs 100%, then AI use should be completely restricted against
But if HN falls into what I call the 2nd category too then does it really matter if the local chef's using a recipe built by chatgpt?
I completely understand what you are saying man and the psychological effects are real.
And this is why I feel like if Hackernews is targetted for people in the 1st category or the 2nd
Because if there is a forum where both local shops who say makes fries which are built in less time and there is also the chef which takes hours building teh perfect food and both of them are on the same forum & they probably get the same space in it then a conflict between them is sort of meant to occur with all the effects which you mention and feels to me like a prisoners dilemma because if artisanal chefs are impacted by local normal cooks lets say then just being a local cook would get stigmatized in a forum. Overall its a net negative for both.
But it honestly depends on how negative if the eyeballs come on both, sure they can split and create different forums but neither wants to leave the large one and the large forum owner has had incentives in both artisanal and local cooks (YC funding AI companies and hackernews once being so damn creative that I hear stories about when one day the moderator decided to have the front page be all about erlang/elixir from a story I remember from a HN "veteran")
I mean I love em both. It's not an or condition but honestly what I hear in HN is the path of least resistance ie. if you don't like moderation, then make your own place but I do feel like the moderation's atleast a bit confused about their interests too.
Probably another reason why I am interested in fediverse because I want it to be like a street where we can have both artisanal chef clubs and the local shops fair while on the same road but a bit of problem with fediverse atleast in this context of HN alike is that lemmy's c/technology can probably replace HN in some contexts but the streets of lemmy are empty because everybody's still here on HN.
I don't disagree with what you are expressing but I think you are overgeneralising.
Sometimes the idea itself can be interesting. If AI is used to manifest the idea then so be it!
Secondly, I think it is almost unavoidable to use some ai generated content (code, design etc) when developing these days. Its just a tool and it can be used well or used poorly.
Lastly, I agree that there does exist a lot of what can truly be labelled "AI slop" where it is just lazy regurgitated content, and that is annoying.
I think there are a few people out there with great ideas, that might benefit from having access to tools to make these real. I'm all for those and would love to see them.
But i also think the majority of these cases will be bad ideas that don't need to be made at all. Or even ideas that are AI generated by asking a LLM for ideas.
I don't care if they are made or not, but i would rather not have those on the internet and especially on 'show and tell' places like ShowHN etc.
I'm also not opposed to using AI as a tool in development or similar.
But there are more and more examples of purely vibe-coded or generated things and those fall in a different category in my opinion.
I am much more interested in infrastructure related things and price optimizing. I've spent hours and days looking at all the options compiling resources and building scrapers to scrape information of servers and having them be made via LLM :) just to get the idea of context.
I don't know but like what i possess isn't a deep knowledge within the domain itself (one of the reasons why I love golang is that all the tradeoffs of it are okay with me and it's complexity just feels linear to me and I would have this simplicity in code where if I ever wanted to, I just feel so much more confident about tweaking LLM generated golang code than say rust)
A lot of what I love is the hacking around the project. Ooh so you used golang to make it cross compatible easier? Are you using the modernc driver? What servers are you gonna use? (hetzner,ovh,netcup etc.) while the code being completely open source and I will be transparent about how much the servers cost and how much I make in the middle and just some sustainable ideals so that it can be day one profitable but without ripping anyone off.
Y'know mixing and matching different services. I think I am really frugal so like this is like the deepest intersection of like all of my hobbies combined.
I face any problem -> use free llm from web -> python/golang code -> deploy it on my web server -> use cloudflare tunnels and custom scripts + zed/others and micro to deploy servers great (I prefer using tmux as well with zed terminal)
But even with all of this, I wouldn't say that the code isn't slop but I do feel like its pretty much reasonable and I have been vocal about how if I ever create something and a community actually occurs out of it & they want me to recreate it from scratch without any LLM assistance, I actually would even if that means reading documentations of all the libraries used in such golang project etc.
But like the reason why I still do things is because I do it for myself and I am pretty impatient. I build things which trouble me first and foremost to fix issues I have & that's kind of source of joy I have.
I don't know man, like I hate AI for all the psychological issues its gonna cause but I still want to build things without having to spend say days in a project when I would be facing the problem right now.
I think I have always imagined myself as some consultant who wants to give the best option out there and have the code back and manage it or similar.
It's just that writing the code part myself isn't the most exciting part to me. seeing computers go work the way I want to in the decision choices that I make and can justify is the most exciting part to me.
Like does this make sense? I am not exactly pro AI either, I have written so extensively about it here on how AI is a bubble and the psychological and social impact it has.
In a way, I don't think I fit in the two boxes of pro AI or completely anti AI. I just am in the middle and I don't know if there's a term for it but like, I don't know if someone knows a term for what I feel about AI it will be brilliant but even if there isn't a mainstream word for it. I just don't really want to change myself to fit in any box. I have my nuanced opinion and I have thought about joining either of these boxes but being honest neither fits my worldview.
> My claim is that they’re seeing something they want to exist and they’re making it exist and putting it out there, while the vast majority of haters aren’t exactly out there contributing to much of anything in terms of “real software engineering.”
Except that they didn't. They thought of something and then asked a tool to make it badly. I know it's hard to separate for a lot of people and it makes them feel like they made something. It's especially bad when that thing then has stuff on it like "Made with attention to accuracy" or some similar marketing claim when there is zero accuracy and a bunch of mistakes in there.
But me running the cmd to create the Hello World angular example does not mean i made anything.
This response makes it 100% clear to me that this is just a bot.
The verbatim use of a completely random thing like 'custom math libraries for cross-CPU determinism', combined with the agreeable tone and em dash use are pretty much a dead giveaway.
The internet has become so useless, everywhere is just marketing nonsense and bots.
If it was for personal use, it wouldn't be on the internet and advertised on sites like this.
So it is at least marketed for personal reasons. I don't know why, but it could be for ad revenue, resume filling, ego bolstering.
This is the problem with AI music generation too. Many people actually feel like they made something and are proud, despite just having clicked a button after asking for a result. This gives them some kind of feeling of accomplishment, despite having done absolutely nothing for it.
The title even says so 'I quit, but AI brought me back'. No, it didn't, you're not back, you just paid someone on Fiver a couple bucks and then used his work to pretend you're back.
This. I have so many things to say about the site, but have been withholding them in fear of "posting shallow dismissals, especially of other people's work"
I will say it for you though. If the calculator doesn't work or shows wrong results, then it is shit. I don't care if made with AI, humans or monkeys.
`Professional Calculators
Free, accurate, and easy-to-use tools for finance, math, and everyday calculations.`
So yeah, fuck this. Website looks 99% similar to the many shadcn themes available, has some errors in the calculations and then have a new account propagating 'ohhh look at me, I did all this with Claude/GPT/my aunt's raspberry pi running X, AI is amazing. All Hail AI.` is tiring and I'm just going to assume it is either paid marketing, people invested directly or indirectly in these companies or just sycophant idiots.
Also, anyone knows an extension or website where I can just filter submissions if they include AI or LLM or Clause in the title? Getting really tired of this shit.
You think it'll rapidly get smarter, but it just recreates things from all the terrible code it was fed. Code and how it is written also rapidly changes these days and LLMs have some trouble drawing lines between versions of things and the changes within them.
Sure, they can compile and test things now, which might make the code work and able to run. The quality of it will be hard to increase without manually controlling and limiting the type of code it 'learns' from.