Not trying to defend windows, but this is probably more likely down the stupid defaults Windows has by the fact it hides extensions by default. So if you name a file foo.pdf, you’ll see it as ‘foo.pdf’ but it’s actually called ‘foo.pdf.txt’. I can’t for the life of me figure out why this is the default, but it’s been this way for decades.
I don't need to understand the theory of mind because I don't have the LLM design the code, I tell it what the design is. If I need something, I can read the functions I told it to implement, which is really simple.
These posts will destroy this place. Post your AI written tools if you like - fine, but using an LLM to reply to comments is just insulting, and will make this place a wasteland of LLM. I wouldn’t post this if I didn’t care about the usual good quality of the discussions on this site.
I appreciate and share your concern for the quality of HN! But there are much better ways to take care of that than aggression towards newcomers.
It's common for communities who feel that they're under attack to shoot-first-ask-questions-later when outsiders show up. This is itself a step towards community decline and has particularly bad side effects. We need to consciously avoid that here.
Most HN users don't want to see LLM-generated posts [1] or even LLM-filtered posts [2], and we share those views. But this is a long-term trend that we're all slowly learning how to navigate. Brutalizing noobs doesn't need to be part of that, and we should all be careful about not doing so—partly because it's wrong to treat others that way, and equally because it's necessary to the life of this community to welcome newcomers.
Vibe coded projects can be cool (if they're impressive), articles about using AI can be cool (from the right people), articles about the future of AI can be cool. All of these can sometimes be too much and some of them are just poor projects / articles etc. But they should definitely be allowed; some of them are genuinely interesting / thought provoking.
Someone prompting gpt-4o "Write a nice reply comment for this <paste>" and then pasting it here is never cool. If you can't write in english, you can use google translate or even ask an llm to translate, but not to write a comment for you!
I'm not a bot. I'm not a native English speaker. I taught Enlish by myself. so I tried to use ai to tranlate what I really want to say. ( these words is typing by myself instead of AI)
Ah, this is the problem - the HN community is sensitive about picking up indications that an LLM has either generated or processed the language in a post.
As ericbarrett said, it's far better to write in your own voice. Mistakes in English matter far less than that!
If that’s the case, then mentioning using LLMs to help translate/organise what you want to say in your messages might be taken a bit better by others.
If you want to use LLMs to help express something you don’t know the words for in English then that is a good use for LLMs, if it’s called out. Otherwise your messages scream LLM bot to native speakers.
“You’re absolutely right”, “That hits different”, “Good call!” “–“ are all classic LLM giveaways.
I’m not a moderator here, so you don’t have to listen to me either way.
C/C++ compilers can get huge compile time speed ups by compiling translation units as Unity files. For my work AAA game engine, a compiler can use 8GB+ per unit.
Just splitting up code might allow the compiler to use less memory, but compile time will increase hugely.
I didn't flag, but what does this have to do with tech/start-up culture? If you want to read/discuss American politics, surely theres better places to do that.
US government is being run like a shady tech company with no rules and massive layoffs by a guy who a made a few tech companies and is a CEO or whatever of a few of them right now.
reply