The point that I've not seen someone making: do you even need LLMs for domestic surveillance? I can grab a copy of EmbeddingGemma or Qwen3-embedding or a similar model and do semantic clustering of existing data, since the "retrieval" is the most important part for such applications, not its integration into a LLM.
If we replaced telemetry with some sort of survey emails and phone calls, we'd get exactly another 500-thread discussion on HN about how "Mozilla is collecting emails to sell to the highest bidder!", "Mozilla is sending us spam!" and whatnot.
Is a migration from language X to Y or refactoring from pattern A to B really the kind of task that makes you look forward to your day when you wake up?
Personally my sweet spot for LLM usage is for such tasks, and they can do a much better job unpacking the prompt and getting it done quickly.
In fact, there's a few codebases at my workplace that are quite shit, and I'm looking forward to make my proposal to refactor these. Prior to LLMs, I'm sure I'd have been laughed off, but now it's much more practical to achieve this.
Right. I had a 100% manual hobby project that did a load of parametric CAD in Python. The problem with sharing this was either actively running a server, trying to port the stack to emscripten including OCCT, or rewriting in JS, something I am only vaguely experienced in.
In ~5 hours of prompting, coding, testing, tweaking, the STL outputs are 1:1 (having the original is essential for this) and it runs entirely locally once the browser has loaded.
I don’t pretend that I’m a frontend developer now but it’s the sort of thing that would have taken me at least days, probably longer if I took the time to learn how each piece worked/fitted together.
A LLM-assisted codebase migration is perhaps one of the better use cases for them, and interestingly the author advocates for a hands-on approach.
Adding the "with help from AI" almost always devolves the discussion from that to "developers must adopt AI or else!" on the one hand and "society is being destroyed by slop!" on the other, so as long as that's not happening I'm not complaining about the editorialized title.
Thank you for posting this. Despite being an old video, I had never come across it, and it almost made me tear up. It showed me the hope that I wished the web would be, despite it never realizing that ambition, with businesses that only pursued engagement metrics, and governments who saw value in vassalizing tech companies to pursue their political goals.
> the signal-to-noise ratio in AI-drafted comms is brutal
This is also the case for AI generated projects btw, the backend projects that I’ve been looking at often contains reimplementations of common functionality that already exists elsewhere, such as in-memory LRU caches when they should have just used a library.
Bad example, you really should just write caching yourself. It’s far too little code to pull in a dependency and if you write it yourself in every project that needs it then you will get good at it, so cache invalidation bugs won’t be an issue.
It wouldn't. If you load configuration and it's the wrong type, now Pydantic is throwing an error. You can't run code if the config is invalid no matter what tools you use: failing validation is still an error.
Back in the day, when the Jacquard machine entered the industry, my advice to weavers would have been to either begin tapping into their network and move to artisanal sales, or to learn how to operate/maintain mechanical looms. The writing was on the wall with the punch card system. In retrospect, it didn't take too long to go from the mechanical loom to diffusion models and world models for robots.
Of course, there was a longer timeline of incremental improvements on the loom over years, so the industry was not transformed overnight. But that's what's worrying people this time. We might not have time to reskill or retrain before financial hardship strikes. Especially in the current political climate. The middle of WWIII is one of the last periods I'd have asked for this automation revolution to take place.
And another practical observation is that not many people have Lobsters account or even heard about it due to that (way less than people who heard about HN). Their "solution" is to make newcomers beg for invites in some chat. Guess what would a motivated malicious actor would do any times required and a regular internet user won't bother? Yeah, that.
I think this is the inevitable reality for future FOSS. Github will be degraded, but any real development will be moved behind closed doors and invite only walls.
reply