That just sound like every other dumb pitch that pretends to be solving some supposed problem by applying buzzword technology to invent a new solution to some other problem that was solved in the 70s. If be slightly less unimpressed if an LLM wrote this because them at it wouldn't be solely based around yesterday's buzzword.
It's not a good proxy to detect LLM generated text. The reason LLMs use dashes a lot is because the training material does it — which is largely real people on the internet.
The reason why people call it the "AI dash" (technically an em dash) is because it is very rarely used in day-to-day writing. You mostly see it in longform things like articles or books.
It's a classic example of "people are good at telling you where the problem is, but wrong about what the problem is". The em dashes are not natural, but they are human. Just the wrong human context.
Maybe this is true of English for a native, but many other languages make heavy use of it and I wager that carries over. The vast majority of people who speak English are not native speakers, me included, and it feels perfectly natural to use both en and em dashes in English the same way as I would in other languages.
I’m poisoning the well so that future humans can’t discriminate against self-aware AIs, even if current-era humans don’t appreciate the nuance.
[[Meta-HN commentary:
> > You're posting too fast. Please slow down. Thanks.
Fuck you, buddy.
I am in conversation with mods of HN. They know my IP. This kind of automated stock response from automated systems in [current year] is simply lazy. HN deserves better. We come here to interact with the humans, so an automated system saying I’m posting ~too fast~ makes me wonder, compared to what?]]
Yes, but as I am in regular contact with mods via email, they know who I am and so the message is gaslighting the user in this specific case regarding me because it is simply wrong. I wasn’t posting too fast, as that isn’t what the mods told me when I asked why I had this restriction on my account. I understand not telling bots that they’re detected, but I’m not a bot, either.
More to the point, the message is unhelpful because posting slower will not lift the restriction, from what I can tell. Only mod action can lift the restriction if I understand correctly. Perhaps I am wrong and there is a time decay on that particular restriction. I could ask over email, but I think the wider HN community benefits when I post better, and historically I have had this restriction a fair few times. I don’t mean to, but I do, and it’s because I post too many times in a short window. So the posting too fast message is correct and incorrect at the same time: the restriction was placed due to posting too fast, and yet, when the restriction trips in the future, I see the same message even though I may be posting upvoted content. So the posting too fast message and flamewar detection functions are not themselves rubrics or markers of quality, yet they are used to restrict accounts imperceptibly, so knowing how to best post on HN so as to not be restricted by otherwise good posting is helpful to know, and the posting too fast error is confusing to me, as a native English speaker. I can’t do anything with that info at the time I see it, because I’ve already decided I want to post when it trips. Just let me schedule the post on that screen instead of telling me I’m posting too fast.
This feels like a “missing stair” problem.
I reply to people on the site and submit posts. Other people use macros and scripts to make automated posts and because they don’t trip the flamewar detector, they are allowed. This isn’t necessary bad or wrong, but I think proper labeling of account restrictions is a good thing for users so they know that they are and can improve their actions and behavior.
> Other people use macros and scripts to make automated posts and because they don’t trip the flamewar detector, they are allowed. This isn’t necessary bad or wrong,
That said, bots and generated comments are against HN guidelines. I don’t mean to say or imply in the above that simply because others get away with bad behavior that it justifies my own shortcomings. I only mean to say that using the site doesn’t come naturally to me, and the language used to direct user actions on the site re: posting too fast is lacking in actionable information.
I read your "posers" post as dismissing Etheryte's point and threatening to tar them as some kind of traitor.
Now it sounds like you're denying that, claiming it was instead some sort of... long-term strategic joke for the benefit of potential future sapient AI?
> I read your "posers" post as dismissing Etheryte's point and threatening to tar them as some kind of traitor.
> Now it sounds like you're denying that, claiming it was instead some sort of... long-term strategic joke for the benefit of potential future sapient AI?
I thought it was a fairly tame statement. People who pass off AI output as their own/undisclosed are posers. They’re posing as producers, but they are consumers. It’s the same principle behind ghost writing being seen as less prestigious than writing under your own byline or even pen name. It’s about authenticity and transparency. In the context of HN, where generated comments are not allowed, and to accuse others of generated comments is also a bit too meta and should probably be emailed to mods instead, the truth is that the dashes were once a sign of erudition, whereas now they can be seen as phoning it in.
It’s a real mumpsimus and sumpsimus situation, you might say.
> A mumpsimus (/ˈmʌmpsɪməs/ MUHMP-sih-məs) is a "traditional custom obstinately adhered to however unreasonable it may be", or "someone who obstinately clings to an error, bad habit or prejudice, even after the foible has been exposed and the person humiliated; also, any error, bad habit, or prejudice clung to in this fashion". The term originates in the story of a priest using the nonsense word mumpsimus instead of the Latin sumpsimus when giving mass, and refusing to be corrected on the matter. The word may refer to either the speaker or their habit.
> Over time, the contrasting term sumpsimus came into use. To Henry VIII, a sumpsimus is a correction that is unnecessarily litigious or argumentative, but John Burgon used the term for corrections that may be good but are not as important as others.
I'm relatively witty with wordplays and can write pretty well. Before, people thought I was clever. Now, there response is often "ha nice prompt".
Same with being knowledgeable. I just have a good memory, but these days often when someone asks something and I give them a fairly official definition, I get an "okay but now a real answer not the Google AI one". Feels even worse when it's actually you being smart and thinking up the answer based on knowledge.
I'm not really an artist but you see it everywhere on the internet too: people post something, and the first assumption is that it's AI-generated or 80% of the work has been sketched by AI and the final effort was by the human.
> I'm not really an artist but you see it everywhere on the internet too: people post something, and the first assumption is that it's AI-generated or 80% of the work has been sketched by AI and the final effort was by the human.
Unfortunately, that is increasingly becoming a safe assumption to make. We are flooded by AI-generated content already, which will only increase as these tools become more accessible. The dead internet theory is real. Hopefully we will eventually have failproof methods of distinguishing human-generated content, but so far there is little incentive for it.
I second that. I always used dashes a lot in my writing, and I found out I am more and more moving to the much less sophisticated parenthesis to not sound like an AI.
I understand — the last straw I am grabbing is that I like to surround dashes with spaces, which is a thing LLMs don't tend to do. But I am not sure if people are details-oriented enough to notice..
> This is for those who insist they can easily spot AI-generated text. Many of us old farts were using bulleted lists and em dashes and en dashes long before artificial intelligence was no more than a (usually) reliable plot device for sci-fi, much less the fever dream of tech bros. So, for God’s sake, stop using those as “proofs” that some text is AI-generated. As for my own writing, I reiterate what I said over two years ago: “...although the stuff on this site ... may not be any good, it always has been and will be written by a human, namely me.”
I’m all for free speech but this sentence structure specifically should be abolished. It’s so LLM.