Yeah, well, I'm tired of luddites like OP. If you want to stick your head in the sand and persuade yourself that the tsunami isn't already upon us, go do it in a private Discord channel instead of whining publicly.
(to be honest, I would consider this the most human quality to create an us vs them dynamic but I meant human as in good)
Why can't there be nuance to the whole situation is something beyond me because in software engineering, from what I am seeing, people have an lack of faith in AI generated/assisted written code and In my honest opinion, its good for early prototypes / seeing the market fit but be honest about what your intentions are
Like personally, I create a lot of software (AI generated) which I use for my own personal enjoyments/problems-issues solving. I generally release them to the public as an after-thought if someone's interested but that's not the reason why I create.
Now tho, if someone is interested in my product or I find a real niche and I find people are willing to pay for the project, I feel like at that point, I will seriously reconsider writing the app myself or weighing in the choices of what to do moving forward but whatever it is, I may do, I will try to do so with full transparency.
As an example of another nuance:- A really good game got disqualified for a game award show because they used AI generated artwork as a filler or something when they were prototyping and then they scrubbed every AI generated stuff from the game when they released to the public.
I see even AI haters defend that it was a good use of AI/understandable use of it so we all definitely need to understand the nuance and have a mature discussion.
Although, in all admittance I am a bit of AI hater myself partially because I see it filled with grifters and even though I use it, Its mostly for my own personal usecases or prototyping purposes mostly and should be treated as such.
In my opinion, using AI generated stuff just generates backlash which just doesnt seem worth it and there are genuine reasons not to use it either like the studies which showed that people became less productive overall with AI and so so many others.
I feel like this is a commentary on the general landscape itself. Like the idea of shipping fast with capitalism feels like an empty promise personally because I'd rather much do with minimalism and KISS policy than adding feature creep.
Sometimes not having some features is a feature in it of itself. Personally I'd take something minimalist designed for the purpose thing which (maybe be prototyped with AI but at the end right now, would be handled by a person) than a thing which will do a 1000 things generated by feature creep with AI
These are something that I grapple with a lot as I really like sveltekit but I hate the node ecosystem and I love golang too but there is definitely a bit more friction in golang compared to sveltekit but the performance upgrades while being cross compatible, simple, mostly very few dependency-needed makes me try to create golang generated websites and so I use LLM's to create golang binaries for personal use for things which I can probably code in a few hours/days with sveltekit
With my focus on using AI to prototype, I feel like its a crutch. Although I only use AI free versions in the web so a more file-based approach like sveltekit definitely has crutches compared to the single file main.go approach that I follow for prototyping.
It would be interesting to discuss this with someone perhaps. I really love golang and I genuinely love it for something minimal but at the same time, for websites, I feel like although I might appreciate golang as an end user, for creating it though, sveltekit is perfect and can be deployed to cf workers too
I have yet to "master" sveltekit but its guides feel like a wonderful thing to go through and although I still ask questions with sveltekit, I am way more involved overall so I am interested in what language I should prototype with.
One can allow massive scalability without resources at a less time being more ergonomic at the cost of overall being harder to grasp compared to the other which might take more resources and might be a bit less ergonomic but moving forward It would be easier for me to grasp
I had always considered memory/resoruces to be expensive (and yeah memory crisis is happening) but with cloud providers like hetzner and so so many others that I got into the rabbit hole of. It genuinely doesn't feel worth the hassle of solely for performance. I genuinely don't feel like It might transition from sveltekit to golang for performance upgrade or god forbid transitioning to rust because even go's garbage collector was too much
Golang feels more simple to view for local code, I can understand things happening in it and I feel like all code is really simple to understand but to take all of the picture in the whole mind becomes complicated as you only rely on stdlib which is good but for websites its really complicated
Whereas Sveltekit feels more magic-y so less localized code overall but overall things do fit nicely (although I still can't code UI myself even with ShadCN for the love of my life but I can use LLM to generate prototypes but it has its own issues)
Sorry for the yap but also not because I am sharing the nuance I face with AI, and so I am interested in asking what are the nuances that hackernews share about AI/ LLM generated code and what are some acceptable policies in your mind? (personally I would strive to one day be on the non AI side than AI)
Like people say that they aren't interested in writing code by hand and I can agree to that since I think we all in the end are motivated by computers going beep boop but at the same time, I have real questioning of ownerships sometimes if I really made a project that I could've made but I instead asked an LLM to prototype for. Should I still create an human prototype of those LLm generated prototypes for learning experiences or that feeling of ownership that I feel so missing at times if I generate a project idea with LLM.