Comment by chrystianpl
3 days ago
As English is my second language and as I have dyslexia. I was just wondering what do you mean by "AI-edited comments"? I can't ask an llm to check if I have made correct grammar/fix it and then as I was on other account, down-voted because of my styling/grammar, not because of the content?
You could always tell your LLMs to just fix your grammar but not embelish, add new ideas, etc..
This is what I do when using AI to read anything I write. Some prompt like "I am going to share with you something I have written and I don't want you to change my voice at all. Can you look for structural issues, grammar or punctuation errors, and things like that". Claude is an amazing editor and I never feel like my writing has been taken from me doing this.
I usually tell it not to rewrite my words, my words are my own. If it has suggestions to tell me what those are, but only fix or show me grammar fixes instead.
Does that work?
It works really well. I've been using this prompt to find spelling and grammar errors for about a year now: https://simonwillison.net/guides/agentic-engineering-pattern...
"fix english" is the prompt i wish to turn into a button
Trust your own style, even if you aren't a native English speaker. Here's an example where a non-native speaker used an LLM to polish his post. The general consensus was that his own writing was preferable to the LLM's edited version.
https://news.ycombinator.com/item?id=45591707
For dyslexia, use a spell-checker. For grammar, use a basic grammar checker, like the kind of grammar checker that has come with MS word since the 1990s. But don't let a style-checker or an LLM rob you of your own voice.
> The general consensus was that his own writing was preferable to the LLM's edited version.
I don't believe a single one of those people.
> For grammar, use a basic grammar checker, like the kind of grammar checker that has come with MS word since the 1990s.
Those are notorious for false-positives, false-negatives, and generally nonsensical advice. Not that the LLM-based alternatives are much better (looking at you, Grammarly), but still.
Oof although I feel this pain a lot. What I like to do is respond to them politely if someone talks about such thing. Although it takes time and this does sometimes make you want to dis-incentivize/dis-engange.
But at some point, the rationale behind it is that your comments are your words and I find it liberating. Some people won't appreciate it and some people would but this goes the same for AI-edited posts too.
(I would recommend to add that if you are still worried, then within your hackernews profile, please talk about you having dyslexia as people might be so much more forgiving when they get more context. We are all humans after all and I would like to think that we understand each other's struggles)
"Please don't post shallow dismissals, especially of other people's work."
I wonder if an explicit expansion of that rule would help. Maybe in all caps. Saying "picking on grammar is a shallow dismissal".
I don't believe that's always true, and I suspect it was left out of the guidelines deliberately, and I wish people receiving suggestions would stop interpreting it that way. Of course people suggesting grammar corrections and treating it like they just demolished and eviscerated your argument are part of the problem. But what about people out here just trying to help? Grammar is important, as it's the syntax of the programming language we all use with each other. People act as if bad grammar is something you're born with, and can't change. Like learning grammar is impossible, and those who don't bother should be a protected class. I'm just trying to help man. Or I was anyway, before I stopped. But if I'm trying to engage with someone's main point, it should be obvious. Whereas a quick grammar correction is just that. But it's a tangent, and not interesting (especially if you already know), and supposedly grammar is "not a technical topic" (despite daily use) so it ends up deemed a "low value comment" and gets downvoted to oblivion.
> I wish people receiving suggestions would stop interpreting it that way
The specific problem here was that the poster was being downvoted for grammar. Of course, that's how he could have read it.
Picking on LLM use is a shallow dismissal, too.
LLM use is what LLMs are best at: spam.
Likewise, I sometimes use https://www.deepl.com/en/write to fix my unidiomatic sentences.
But I can see why the HN guideline is formulated that way. My students often use the excuse "I did not use AI for writing! I wrote it myself! I only used AI to translate it!" Simply disallowing all kinds of AI usage is much easier than discussing for the thousandth time whether the student actually understands what they have written.
Isn't the whole point to understand? If the task is to write and you expect only final result, but you question if it looks legit enough, how is it fair judgement? People can deliver partial results and show progress as well, but you won't see it in some comments on the internet, but if something is expected to take many days, it's easy to show different stages of work. It's easy to accuse people of plagiarism or not thinking for themselves, and of course there are indicators when someone uses AI, but the problem is that you can't distinguish in reliable way, if something was created by AI or not.
Like, there is this computer game, authors used some models or something like that, generated by AI, but it was only used during prototyping and later it was replaced by proper models. No one would know about that, if authors would not tell about it. So, if someone writes in their own words what AI generated for him, is it still argument made by human or by AI? What if someone uses AI only as placeholder and replaces all that content, so you never actually see actual AI usage, but it was used in the process?
For me, premise that using AI in any form invalidates your work, starts with logical fallacy, so such arguments against using AI are weak. It's like saying that your work is wrong, because you used calculator, so your calculations can't be right, if done by machine, because it had to make mistake or that's wrong for ethical reasons or whatever.
Work generated by AI can be easily poor, because these models make mistakes and like to repeat in certain ways, but is it wrong that I'm writing comment with keyboard, instead of writing letters with pen? Is it wrong, when I use IDE or some CLI to write code with AI, instead of using vim and typing everything on my own? Is it wrong that someone uses spell-checking?
In the end it doesn't matter who seems smarter, when you're expected to use AI at work. Reality shows you actual expectations.
I am not saying that completely disallowing AI is the right decision. But if you see text that is clearly generated by AI and does not make any sense, it sure would be nice if you could just tell the students to actually read their sources instead of having to argue with them why they should do so. Similarly, I can see why HN moderators do not want to argue with the 100s of spam posters per day on /newest.
Anyway, my university did not ban AI, and now most students have degraded to proxies between teaching assistants and ChatGPT.
1 reply →
I don't have dyslexia but I feel your pain. I mean it is what it is. I would rather have it raw then have to use AI to filter to comments that make sense.
Dyslexia was my first thought as well. The intent is great, but I don't know if this is keeping with the social model of disability. Disability is created when you remove access and this is exactly that.
The internet has been full of brilliant dyslexics since the start, just as it has been full of brilliant blind people. Dyslexic people feeling that they must use AI to produce perfect prose lest they burden the lexics with clumsy spelling or grammar is far more hostile. We didn’t have slop machines 5 years ago.
> The internet has been full of brilliant dyslexics since the start
And they've been nitpicked to death for just as long. Now they have better tools to preempt that nitpicking, only to now be nitpicked over choosing to use those tools. Go figure.
I don't really see the issue, as long as there's human thought behind whatever anyone posts. It's frustrating to argue against someone that lazily uses AI, but if argument is fair, then I don't care if that's written by AI or human, what difference does it make? It's frustrating, if someone is incoherent and makes dumb argument, but again, I don't care if it's dumb argument from human or machine.
For me it sounds just as yet another form of gatekeeping, so either you sound human or you're not good enough to post/comment. Like, really? How isn't that genetic fallacy? It doesn't matter what someone thinks, because someone used AI to make their thought clearer, so their whole argument is trash? Like it has to hurt to read and write, if you're not using English perfectly and your work is seen as inferior based on superficial factors like proper grammar and style?
It's dumb crusade, I did not use AI to write this comment, but I hate when people try to monopolize the truth and tell who is "better, smarter" based on irrelevant facts. Not using AI doesn't make anyone superior. Using AI also doesn't make you superior. Focus on what you mean, because that's what matters.
I don't see how you can know why you were downvoted. Even if one person says something, they won't all. Your comment right here has some rough patches, but I can tell what you're saying. Humans are terrific at extracting signal from noise. I would say be who you are, tough as it may be, and it'll encourage the rest of the world in the future to do the same. We're all unique in some way or another and have flaws and we'd be better off if we were knew others had them too because they weren't constantly trying to hide it and we wouldn't feel so bad thinking we're the only ones. I hope it doesn't sound unsympathetic. I understand where you're coming from intellectually, but don't have any real experience being ridiculed or bullied. I know kids can be brutal and probably scarred you, and unfortunately, adults aren't much better, but we should be, and I think at least Hacker News is better than most places full of human adults. We know there's a huge world out there. I think I'm reasonable well-spoken in English but can't speak a lick of any other language at all. The fact that you can produce intelligible English already puts you above me in my book. You're a person. I can respect you, esteem you, potentially love you, not in spite of your flaws, but because they don't matter. Every single person on the planet has them, and if they're not moral flaws, nobody should give a shit. I can't respect or love a machine any more than I can a rock. And I don't want to talk to one, either.
No worries, it’s unenforceable.
How do you know what you were downvoted for?
I guess he was told because otherwise you don't know whether you said something inherently wrong or misleading or you hurt someone 's feeling.
That's the richness behind the upvote/downvote that also tend to create echo chambers because you soon learn what causes downvotes.
I've personally noticed downvote whenever I mentioned apple negatively.
I have never downvoted for this, and I hope no one else would do that either. If anyone here does that, please stop.
[dead]
[flagged]
> boooooooo, hu, baby
> stump along, cut your own path, or fuck right off
> real life will eat you otherwise
> I mean holly shit, you actualy want to hide behind an automated echoing device so that you wont get, well, what is happening to my post as sooooon as I press↓
You deserve a ban for this.