561–570 of 1001 posts
Good. This helps establish it in the HN culture. That’s the purpose of guidelines. 99% of rule enforcement, both IRL and online, comes down to individuals accepting the culture. Rules aren’t really for adversaries, they are for ordinary situations. Adversaries are dealt with differently.
This is only meaningful if enough people read it and agree
There should be a "flag as AI" link in addition to "flag" and then a setting for people to show flagged as AI. Once the flagged as AI reaches a certain threshold then it disappears unless you enable "Show AI". Maybe once enough posts have been flagged like that then that corpus could be used to train an AI to automatically detect content generated by AI. That would be cool. Maybe the HN site wouldn't add this feature…
I’ve actually been thinking about this exact idea for https://hcker.news/ . Stay tuned, I’ve already started rolling out some comment filtering.
"AI-edited comments" is a very interesting one. Where is the line between a spelling/grammar/tone checker like Grammarly, that at minimum use N-Grams behind the scenes, and something that is "AI" edited? What I am asking is, is "AI" in this context fully featured LLMs, or anything that improves communication via an automated system. I think many people have used these "advanced" spellcheckers for years before Chatgpt…
You should use your own words. It might seem that a tool like Grammarly is just an advanced spellcheck, but what it's really doing is replacing your personal style of writing with its own. It's better to communicate as an individual, warts and all, than to replace your expression with a sanitized one just because it seems "better." Language is an incredibly nuanced thing, it's best for people's own thoughts to come t…
"Your unique human voice is more valuable than a thousand prompt-driven LLM doggerels."
Consequently, I hardly ever spend the time to write out long and detailed HN comments like I used to in the pre-LLM era. People nowadays have a much harder time believing that an Internet stranger is meticulously crafting a detailed and grammatically-airtight message to another Internet stranger without AI assistance.
All the weak excuses posted here are just making me lean more towards a hardline policy. No I don't want to read a human-generated summary of your llm brainstorming session. No I don't want to read human-written text with wording changes suggested by an llm. No I don't want to read an excerpt from llm output even if you correctly attribute it. I acknowledge this is partly just my personal bias, in some cases really n…
I feel you. I don't think I've ever finished reading a sentence that started with "I asked and he said..."
An alternative I tried was sharing links my LLM prompts/responses. That failed badly.
I like the parallel with linking to a Google/DuckDuckGo search term which is useful when done judiciously.
Creating a good prompt takes intelligence, just as crafting good search keywords does (+operators).
I felt that the resulting downvotes reflected an antipathy towards LLMs and the lack of taste of using an LLM.
The problem was that the messengers got shot (me and the LLM), even though the message of obscure facts was useful and interesting.
I've now noticed that the links to the published LLM results have rotted. It isn't a permanent record of the prompt or the response. Disclaimer: I avoid using AI, except for smarter search.
The rule has been around for years, but only in case law, i.e. moderation comments ( https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que... ). What's new is that we promoted it to the guidelines. Fortunately I found some things we could cut as well, so https://news.ycombinator.com/newsguidelines.html actually got shorter. --- Edit: here are the bits I cut: Videos of pratfalls or disasters, or cute animal pic…
I'd be a wee bit cautious with the "AI edited" part of it; since that might exclude a number of people with disabilities or for whom english is a second (or third, or later) language. My reading is that the intent is to have a human voice behind the text. Monitor and see how it goes I guess!
Also writing a draft in Google Docs and accepting most [2] of the corrections is fine. The browser fix the orthography, but I 30% of the time forget to add the s to the verbs. For preposition, I roll a D20 and hope the best.
I'm not sure if these are expert systems, LLM, or pingeonware.
But I don't like when someone use a a LLM to rewrite the draft to make it more professional. It kills the personality of the author and may hallucinate details. It's also difficult to know how much of the post is written was the author and how much autocompleted by the AI:
[1] Remember to check that the technical terms are correctly translated. It used to be bad, but it's quite good now.
[2] most, not all. Sometimes the corrections are wrong.
I'm absolutely 100% for this policy. My only caution is that good writers and LLMs look very similar, because LLMs were trained on a corpus of good writers. Good writers use semicolons and em-dashes. Sometimes we used bulleted lists or Oxford commas. So we should make sure to follow that other HN rule, and assume the person on the other end is a good faith actor, and be cautious about accusing someone of using AI. (I…
> Good writers use semicolons and em-dashes I use semicolons a lot. If this is the nouveau tell du jour for LLMs then I'm in trouble.
At the end of the day, I'm here because of all the thoughtful commenters and people sharing interesting stories.