Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

We're losing our voice to LLMs

tonyalicea.dev

Post reply on HN
Print view View original

361–370 of 425 posts

Re: We're losing our voice to LLMs

#361

Post by Spivak »

hhh wrote: post #359
gordonhart wrote: post #250

Earlier quoted context omitted.

My theory is that LLMs are accelerating [online] radicalization by commoditizing bland, HR-approved opinions. If you want to sound like a human on the internet, for better or for worse the easiest way is to say something that would make Anthropic’s safety team have a heart attack.

Bullshit. You can just use ANY abliterated model or non-safety aligned model. People have popularized ‘hard r or you’re hardware’ for a while now, but any self hosted LLM that’s de-aligned will gladly say whatever. People believe this and continue to get fooled by LLMs all day.

I don't think I would want to hang out in any online space where the n-word is a shibboleth. The practice of asking members to do something that would ostracize you from the rest of society is cult behavior—even if it's not intended as such.

Re: We're losing our voice to LLMs

#363

Post by curioussquirrel »

tptacek wrote: post #274

If you care about voice, you still can get a lot of value from LLMs. You just have to be careful not to use a single word they generate. I've had a lot of luck using GPT5 to interrogate my own writing. A prompt I use (there are certainly better ones): "I'm an editor considering a submitted piece for a publication {describe audience here}. Is this piece worth the effort I'll need to put in, and how far will I need to…

+1 on this one! I only use LLMs once I'm done with writing, and basically using them as my editor.

In case it helps anyone, here is my prompt:

"You are a professional writer and editor with many years of experience. Your task is to provide writing feedback, point out issues and suggest corrections. You do not use flattery. You are matter of fact. You don't completely rewrite the text unless it is absolutely necessary - instead you try to retain the original voice and style. You focus on grammar, flow and naturalness. You are welcome to provide advice changing the content, but only do that in important cases.

If the text is longer, you provide your feedback in chunks by paragraph or other logical elements.

Do not provide false praise, be honest and feel free to point out any issues."

(Yes, you kind of need to repeat you're actively not looking for a pat on the back, otherwise it keeps telling you how brilliant your writing is instead of giving useful advice.)

Re: We're losing our voice to LLMs

#364

Post by rcxdude »

drbojingle wrote: post #354
rcxdude wrote: post #155

Earlier quoted context omitted.

IMO there should not be an algorithm. You should just get what you have subscribed to, with whatever filters you have defined. There are better and worse algorithms but I think the meat of the rot is the expectation of an algorithm determining 90% of what you see.

Dude that's an algorithm.

Not in the sense that it's commonly used in this context. It's not a recommendation algorithm pulling from the whole platform based on what you're doing, it's far more controllable, deterministic process which only does what you explicitly request.

Re: We're losing our voice to LLMs

#365

Post by Genego »

This will get better over time, there will be ways to make LLM output more uniquely yours. I personally enjoy writing all my blog posts with LLMs, because its the only way I can bring to turn countless of notes and drafts I have while running experiments, into some public facing documentation and blog posts that I would want people to read. There are at least 5-6 years of lost ideas, thoughts and notes that I was not able to communicate the same way I can do now with LLMs. So I definitely found my voice here.

Re: We're losing our voice to LLMs

#367

Post by anonymouskimmer »

mold_aid wrote: post #356
anonymouskimmer wrote: post #234

Earlier quoted context omitted.

I'm taking "same way" to be read as "authoritative", whether de facto or de jure. Basically by dint of people using what's provided instead of coming up with their own. Everyone has the same few dictionary spellings (that are now programmed into our computers). Even worse (from a heterogeneity perspective), everyone also has the same few grammar books. As examples: How often do you see American English users write "c…

Right, prior to "dictionary-ization" you could have variant spellings etc. You're correct about prescription. It's an interesting idea, how to understand authority in these contexts. Transitions in media and writing - you could take Charlemagne's attempts to standardize HRE script in the administrative state as a sort of "standardization of Empire," the development of English as the web's lingua franca, the role of t…

And just as important, we don't want people unable to understand things that aren't presented in one particular way. Something we're seeing with attention-grabbing short-format works that leave people unable to muster the energy to stick with something for 5 pages or an hour of video.

It's one thing to not be able to formulate a convincing argument without resorting to the 1-3-1 style. It's another thing entirely to not grasp an argument formatted in a different style.

Re: We're losing our voice to LLMs

#368

Post by lupire »

Spivak wrote: post #361
hhh wrote: post #359

Earlier quoted context omitted.

Bullshit. You can just use ANY abliterated model or non-safety aligned model. People have popularized ‘hard r or you’re hardware’ for a while now, but any self hosted LLM that’s de-aligned will gladly say whatever. People believe this and continue to get fooled by LLMs all day.

I don't think I would want to hang out in any online space where the n-word is a shibboleth. The practice of asking members to do something that would ostracize you from the rest of society is cult behavior—even if it's not intended as such.

Censoring the n-word is an even worse shibboleth.

https://iowacapitaldispatch.com/2025/11/13/school-fired-work...

Re: We're losing our voice to LLMs

#369

Post by lupire »

tptacek wrote: post #274

If you care about voice, you still can get a lot of value from LLMs. You just have to be careful not to use a single word they generate. I've had a lot of luck using GPT5 to interrogate my own writing. A prompt I use (there are certainly better ones): "I'm an editor considering a submitted piece for a publication {describe audience here}. Is this piece worth the effort I'll need to put in, and how far will I need to…

I simply tell the LLM to call out my mistakes and explain them, but do not offer corrections or replacements. I use it to help my kids with their homework and it's fantastic.

Re: We're losing our voice to LLMs

#370

Post by lupire »

khannn wrote: post #296
xeonmc wrote: post #264

Earlier quoted context omitted.

Itemized Bill: - hitting the ChatGPT button: $1 - knowing where to hit the ChatGPT button: $600k/year

I know some GPs that use WebMD so it plays out like: - Google search: free - Having the schooling, training, and experience to evaluate the results: $300k per year

- Google search: free

- Having the license to make it "official": $300k per year

Post reply on HN