Hacker News Reader

Comment by why_at | Hacker News Reader

I can appreciate LLMs for some use cases, but writing emails for the user is the one that really baffles me.

It's one thing if you don't speak English well and could use some help making yourself understood, but the amount of native speakers using this is so strange to me. How does this help you? If you can write to the LLM telling it what kind of email to write, you might as well just write the email.

View on HN

The most frustrating thing to me is to receive a 5-paragraph-plus email that was clearly written with some AI that filled in the email with vapid and useless talking points, like "Let me know if you need any other blah blah blah; While there is clearly a need for system improvement, we are working hard to address the underlying and fundamental issue; This is a lesson that it's not just a feature, it's a critical path for our users, etc."

My theory is that people are fundamentally averse to the thought and effort it takes to write a good quality email. Then there’s probably some underlying belief that more volume shows more effort, which people will perceive positively. And finally, there's the worry that if you write the email yourself, you might make some embarrassing wording, grammar, or spelling mistake.

Specifically, lets imagine LLMs as compilers - you're passing your prompt through to get some pretty language at the end.

Don't send me your compiled code, send me your prompt. Let it be rude, if the wording is awkward I guarantee I can understand it just as well as an LLM, ignore the fact that my daughter just graduated and offering hallucinated platitudes.

Send me the actual question, don't make me try and decompile a big blob of empty text to the ten word prompt that contains all the actual meaning.

AI is a useful tool for a variety of purposes, what it is not useful for is expanding a short statement to an essay and reducing an essay to a short statement. Either the communication deserves to be an essay or ten words will get it done.

Just because you yourself are OK with being talked to rudely, doesn't mean others are. In fact I'd wager most aren't consciously or unconsciously...

Yeah that attitude will not get you far in life unless you're Steve Jobs, and it'll sink your ship unless you're obnoxiously rich. And even if you're either/both of those things: A. you can and should act better, and B. people will always attach an asshole-asterisk to your name for the rest of your life and probably even a good while after.

Exactly. In my 25+ year career, I've encountered maybe two dozen or so people whose e-mails and chats were terse, yet admittedly succinct, one-liners and most of them were also raging assholes to work with. The ones who also didn't use capital letters or punctuation in their communications were uniformly assholes.

My experience is completely opposite. Majority of the people send short emails - sometimes badly written, sometimes well written. And when someone sends long emails, people dont read them. They kinda skim them and ignore the rest.

And no, they are not all raging assholes.

Correct, I was referring to being arrogantly blunt and insensitive, not short/efficient emails.

I'm pretty sure "rude" in this context means "brief and to the point", not insulting. Otherwise you can be rude with an LLM as well.

Most people I know are happy to receive a focused email rather than an LLM-enhanced, 6 paragraph wall of text.

But it's far less rude to just bluntly say something than to send an email generated by an LLM.

I think rude was the wrong word to use. I more meant lacking the pomp and circumstance fluff. I always appreciate considerate and polite speech and think it's requisite to being taken seriously. However, I think directness within the bounds of politeness is optimal.

Also, if it's wall-o-text or "staging must be updated before our os version is deprecated sunday" I prefer the latter.

yeah this is what drives me crazy about LLM writing. Most of the time the prompt has all the info you need and is like maybe a few sentences. Then the LLM expands it into a few paragraphs...

I guess if someone is writing like a big fancy email to send out in bulk, maybe using an LLM to improve would make sense... but just emailing some coworkers it seems super lazy and insulting to send an LLM output :-I

Some people send LLM generated replies in Slack chats! Now there's that.

I wonder if slight grammar incorrectness (like not capitalizing your sentences or using abbreviations) is going to start becoming a signal of authenticity for people subconsciously. Maybe it already has.

if you are prompting such that the LLM isn't pulling context that the recipient doesn't have access to, then your email is likely marginal.

ie the prompt "Send 'bob' an email with a description of why the VPN bridge isn't working so they can debug their side" is a mostly useless as a prompt for anyone, it's only useful when the LLM has all the context of some analysis of the particular issue and what is going on and then injects it into the email.

Does the recipient already have that context? If you want to share some internal context (say you're a front-end specialist and the recipient is quite unfamiliar with the limitations of your framework) then maybe that'd be helpful? If it's just regurgitating already communicated information then either

1. You are restating the information because you don't believe the recipient understood it the first time and thus you should be very precise in your expression to make sure it isn't too arcane for them 2. The recipient appears to understand the information already in which case why restate it?

Either rely upon everyone else changing their behavior, or give up and use your LLM to re-compress incoming messages to be informationally dense as you see fit.

It's lossy tho. LLMs are crap at picking the "good stuff". Eg: the summary of the email covered the point about the family event but missed that the deal-terms were moving from Wyoming to Delaware.

Personally, I'm confident with my level of output so I'll continue to dutifully read all the crap that gets sent to me on company time. I'll just prefer to engage with people who communicate well and encourage that in others.

A privileged position to be in, for sure! Congrats

I prefer not communicating with people who don't respect my time.

Tangentially, yes, let's imagine LLMs as compilers.

How insane is it to advocate the usage of these non-deterministic compilers, where each time you compile may produce different semantics?

And then people resort to saving and hand-editing the compiled output.

But when they want to change the source, they recompile and have to start over hand-editing the output again.

Actually, in professional usage in a technical setting this is my prime objection to heavily LLM driven development. Were the tools in usage deterministic then I'd be a lot less objecting to the mandating of their incorporation into workflows.

I want to be reading, writing, testing and maintaining the software at the same layer. Right now extreme AI usage leads to reading, testing and maintaining happening in a less expressive language than writing and guess which of those four activities developers enjoy the least and find the hardest - it sure isn't writing.

More than non deterministic : LLMs don't have a specification to obey to in the first place, while compilers (rather, programming languages) do.

> Then there’s probably some underlying belief that more volume shows more effort

Hopefully, LLMs will kill that attitude in the long run

They're already making it worse ...

LLM generated texts are an API ideal for interfacing with boomers

The boomers I know are grumpy, impatient, can smell bullshit a mile away and are almost insultingly terse. Which is honestly refreshing.

So LLMs have no place for me in this regard.

Yet they immediately send their life savings to India when contacted

> receive a 5-paragraph-plus email that was clearly written with some AI

Which is promptly and ironically summarized by AI on the receiving side

It's almost as though someone was put in charge of AI growth and all they care about is token burn.

thats completely illogical and no business would ever waste their money something like that (eyes-looking emoji)

O ho ho - [hearty laugh]

Sure, but LLMs are inherently lossy. There is no guaranteed way for the second AI to extract the original prompt from the text.

LLM-written emails are too wordy. But maybe people think that’s what a good email is.

(Did an LLM write your post?)

I would never disgrace you with reading my LLM output unless I explicitly identified it

> more volume shows more effort

I didn't have time to write a short letter, so I wrote a long one instead. - M. Twain

It wasn’t Twain who wrote that letter: that quip was posthumously attributed to him, and have been used by multiple others.

The earliest source we have for it is a letter by Blaise Pascal, some 250 before Twain ever thought about writing letters, or anything else for that matter.

TIL. Thanks

It seems to me like adoption of AI in the workplace has become mostly 1) authorship of verbose emails and 2) summarizing verbose emails for non-tech industries

Don’t forget:

- write verbose tickets

- summaries the verbose tickets

Eh, well, I have been trained by IT staff long before to write very detailed explanations of issues, attempts at fixing, step by step reproduction instructions, include screenshots, etc. I am sure AI has elevated it but the summarizing nature is probably a net benefit for those reading/managing tickets. I’m not in a ticket queue position other than as an end user needing support, so my only data point I have on it is the desktop support guy in my office says he’s on board with it all simply because he can write his snarky comments as he’s always wanted to and then just tell the LLM to make it friendly before sending.

Well, business english IS annoyingly verbose and full of empty phrases. It would be cool if we could dispense with vapid pleasantries.

I'm certainly not going to be the first to stake my job or my promotion on that particular hill. So I can fully understand why people will still turn "We need the database changes ASAP, you promised they'd be done. Get it done!" into half a page of drivel.

It’s the protocol of the brave new world, you and the recipient need a single sentence to communicate but the culture dictates using certain language and politeness + personal flavor so your AI helps you write culturally appropriate fluff and the person who receives it is using their AI to get rid if the fluff so you are both optimized for productivity through stripping the culture away making your interactions faceless and yourself fungible.

You can imagine this spread into dating as well, so you just have sex efficiently to optimize the breeding and hedonism.

At some point the protocol of expanding and then compacting with AI will also be removed to optimize the unneeded inference and people will again talk to each other but using the caveman language, stripped away from centuries of culture.

My ADHD rejects modernity. I shall type novels when engaged in discussions about feature design decisions and if your question has an easy answer I will give it to you shortly.

I absolutely agree with your opinion and I loathe it.

I (mostly) welcome _people_ writing medium to long form emails, slack posts, and pull request comment (and the like).

What really grinds my gears is when the clearly desired response is a few words or a single sentence, and what I get is a link to an obviously llm generated 3 page pdf full of em-dashes and emoji bullet pointed lists with very little relevance or context about the question.

If I wanted Claude or ChatGPT's response, I would have asked them. If I'm going to bother a cow orker with a question, it's because I want domain specific knowledge or workplace experience that might be important.

I'm more and more often internally reacting with "if you didn't bother writing it, I don't need to bother reading it".

I would welcome your 30 min or half day turnaround with a well written and thought response, over lazy/disrespectful colleges who are just doing the instant 2026 version of "just fucking google it".

Hey, speaking of gear grinding - have you run into LLM generated comments? I have always loathed JavaDoc and friends as I think it encourages vapid space filling comments that are inherently knowable from the code - when it's connected to a renderer (as was the original JavaDoc) so that the comments can be exposed without the code that is fine-ish - it serves a purpose and I can comprehend the rational but in most cases I've seen those comments committed without any intent to ever render them separately.

In the modern world we've got comments written by LLMs because "You've got to write a comment, of course, it's required!" but now the actually significant comments (the Why comments - as opposed to the What ones) are lost in a sea of LLM slop so no one will read them. Considering it'd be just as easy for the reader to point a conversational LLM at the codefile if they want the LLM interpretation of what's happening why are we bothering committing it at all?

Gosh that really grinds my gears. It's definitely a tangent but that being encouraged is a huge red flag for me.

Couldn't figure out if this reply was serious or sarcastic. If the latter, then it's hilarious.

Youtube implemented the same sort of thing for channels. If you have a youtube channel and someone comments on one of your videos, there is an AI-generated "reply" that you can click to avoid having to actually think about interacting with commenters on your videos.

The weird thing is, if I commented on a channel and they sent me an AI-generated reply, I'd just hate them forever.

It's insecurity. They worry they might be saying something dumb and the LLM gives them assurance that it sounds "better" and "more professional."

And they will just get worse at writing anything by depending on it. Soon enough practically nobody will be able to write.

It’s a false assurance. LLMs say dumb things all the time.

Ronny Chieng's speech at Harvard's Class Day that went viral put it well, something along the lines of "AI can write emails and summarize threads for you. You know who else can do that? Me."

It is no surprise to me at all that some people reach to bots for help with writing email.

I've seen some very incomprehensibly-written communication in my time, including from people who speak English as a first language.

The most frustrating group of consistent offenders I've seen was comprised of folks who absolutely should know better: School teachers.

21% of Americans are functionally illiterate. 54% of Americans read below a 6th grade level.

Things that seem trivial to someone well-read can be insurmountable tasks for millions of people.

Those people may do fine day to day, you don't need to write all that many emails for many if not most jobs, but when they do need to get their intentions across, things might get dicy.

Or course this solution is far from perfect. If you use LLMs to word your message because you're not very proficient in reading or writing, you may not understand what the output is saying, so you can't verify that it didn't misinterpret you.

You have to be able to write enough to prompt an LLM. So why not just send the prompt, since that's what you actually said?

Gmail can guess what you want to say without you having to type a prompt.

well what's the point of even sending the email if gmail can just reply for you

I don't know lol but apparently some people use that function. It doesn't make any sense to me

Because it's already hard enough to get someone to understand what you're saying if you write full, flowing sentences.

Back when I did customer support, I occasionally dealt with people who lacked the capability to express themselves in written language. These weren't people born in other countries or people with some kind of intellectual disability, often they were just never taught how to read/write/compose a message.

There's also a social stigma to sending a job application written like "good. morning. sir i. wnt too. Wrk for ur Compny u. hve job. for Me im steven i.m" and so on, even if the job in question has nothing to do with any form of written communication.

Plus, there are plenty of things people just don't know how to do. I once had to Google "how do I write a letter of resignation" and Google will now generate a fulll message for you if you type "i want quit next month how write".

I think the author of this blog post assuming a prompt to help improve a message (obnoxious glowing button aside) somehow implies that Google thinks they're stupid showcases the perceived link between the ability to write and intelligence.

Obviously, using LLMs for this is just patching over the lack of resources put into (adult) education and literacy, as well as a societal shift to stop treating illiterate people badly. However, structural improvements take many years, probably several rounds of elections, and doesn't help someone today.

I don't use gmail but often get an LLM to write certain emails. The benefit is that it can pull in context and typically one-shot the email without me prompting it at all.

For example, a tenant emails me about some issue relating to a specific property. It can go through my leases, find the right one, check other emails to see I ordered a new appliance to that specific address, track shipping/install, all that, then reply appropriately.

I think your example is just customer support and not something that requires a personal touch anyway. Like your tenant doesn't care about your tone in that context, just the information. (It does seem like email is the sub optimal channel for this task anyway).

But if you're writing to someone with the intention of communicating personally, using AI anyway shows a lack of effort.

More pointedly, instead of prompting the AI to write something, just send the prompt. It will be concise and pointed, exactly what a message should be.

> It's one thing if you don't speak English well and could use some help making yourself understood

If someone reading does this, please, do not. Imperfect English is a lot more pleasant to read than AI slop. It will not sound better, it will sound worse.

Everyone says so, and then when you make even a single mistake in grammar, orthography or use weird phrasing, your entire point is disregarded.

And currently many phrases we've been taught to use for formality are being seen as a sign of LLM usage...

I think it's a case of projection. Megatech CEOs feel that avoiding interactions with peons is the most important thing, so they add this feature assuming that everyone else also wants that.

speaking is faster than typing. My current process is:

1. Use Google Eloquent or directly into ChatGPT to dictate the email.

2. Then ask AI to polish, simplify, summarize, and I provide my writing style [0].

That gets sent to the user. I am very careful to not allow the AI to write too much.

The alternative is I spend 20+ minutes writing and re-writing the email.

[0] - https://www.kcoleman.me/writing/slack/2023/03/11/writing-sty...

As an aside, using an LLM results in a post that's all black text, except a couple of colorful emojis that jump out:

> Speak with confidence. “I think that I can possibly consider agreeing” . Don’t share weak decisions or opinions. Either agree, or disagree.

When I see those I mentally downgrade the whole text.

Normies love this shit because it makes them fit in the crowd effortlessly. Same reason why corporate slop was a thing even before AI.

Personally, when I message people I respect I either don't use AI or ask it "please fix typos only", but if it's someone I don't give a fuck about, then AI-generated slop it is, because assuming that the recipient is a random person, AI-generated slop has the highest chance of actually getting shit done.

Honest answer, I’m sort of not neurotypical, not formally diagnosed but enough people in my life have gone out of their way to inform me, and so while I do still write emails myself, if it’s for work I usually dump my email into an LLM before sending it and just ask for some minor edits where I could be misconstrued as being rude or offensive. It’s stuff I would normally not notice, but other people take issue with my overtly direct nature of speaking.

Wife is a GP and has to frequently write emails to either insurances or lawyers/judges. All done in french which is a messy language full of obsolete over-the-top completely dishonest empty fluff, but native speakers tend to get quickly offended if its missing, done by non-native speaker. Mistakes can easily have legal consequences.

She uses chatgpt for such with appropriate checks afterwards, all the time. It saves her tons of time, and time is money, time is life spent with kids instead of useless bureaucratic work.

I have tons of similar stories, friends (surgeons, radiologists, all non-native speakers) using chatgpt to write some official application for work permits, motivation letters etc. Just because you don't have certain way of working or work situation doesn't mean its valid for all of us.

I think you might be suffering from a little bit of bias concerning your own comfort with written english with regard to the general population's comfort with written english.

I had a high school teacher (algebra II), my favorite coincidentally, who was actually functionally illiterate. He knew some words, and had a solid understanding of the alphabet, but ask him to parse a sentence or god forbid an essay, and he was completely lost.

He was a native born american, english as only language, and simply could not interact with the written language. If he could have written (in 2009) "give 20 question test for week 1 algebra II student with answer guide" he'd have gotten (based on the prompt I just submitted):

Week 1 — Algebra II: 20 Practice Questions

    Simplify: 3(x + 4) − 2(2x − 5).
    Solve for x: 5x − 7 = 2x + 11.
    Solve: 2(x − 3) = 3(x + 1).
    Solve and check: (x/4) + 5 = 11.
    Factor: x^2 + 5x + 6.
    Factor: 4x^2 − 9.
    Factor completely: x^2 − 6x + 9.
    Multiply and simplify: (2x − 3)(x + 4).
    Expand: (x + 2)^2.
    Solve quadratic by factoring: x^2 − x − 12 = 0.
    Use the quadratic formula to solve: x^2 + 4x + 1 = 0.
    Simplify: (3x^2y)(2xy^3).
    Simplify: (x^5)/(x^2).
    Solve for x: 2^(x+1) = 16.
    Evaluate: f(x) = 2x^2 − 3x + 1; find f(2).
    Write equation of a line in slope-intercept form with slope 3 and y-intercept −2.
    Find slope of the line through points (1, 4) and (5, −2).
    Solve the system by substitution: y = 2x + 1 and 3x − y = 4.
    Solve the system by elimination: 2x + 3y = 7 and 4x − 3y = 5.
    Simplify and write in simplest radical form: sqrt(50).

Answer Key

    3x + 12 − 4x + 10 = −x + 22
    5x − 2x = 11 + 7 → 3x = 18 → x = 6
    2x − 6 = 3x + 3 → −6 − 3 = x → x = −9
    x/4 = 6 → x = 24
    (x + 2)(x + 3)
    (2x − 3)(2x + 3)
    (x − 3)^2
    2x^2 + 8x − 3x − 12 = 2x^2 + 5x − 12
    x^2 + 4x + 4
    (x − 4)(x + 3) = 0 → x = 4 or x = −3
    x = [−4 ± sqrt(16 − 4)]/2 = [−4 ± sqrt(12)]/2 = [−4 ± 2√3]/2 = −2 ± √3
    6x^3y^4
    x^3 (assuming x ≠ 0)
    2^(x+1) = 16 = 2^4 → x + 1 = 4 → x = 3
    f(2) = 2(4) − 3(2) + 1 = 8 − 6 + 1 = 3
    y = 3x − 2
    slope = (−2 − 4)/(5 − 1) = (−6)/4 = −3/2
    Substitute y: 3x − (2x + 1) = 4 → 3x − 2x − 1 = 4 → x = 5 → y = 2(5)+1 = 11
    Add equations: (2x+3y)+(4x−3y)=7+5 → 6x = 12 → x = 2. Then 2(2)+3y=7 → 4+3y=7 → 3y=3 → y=1
    sqrt(50) = sqrt(25·2) = 5√2

I'm sure there are many more accesibility stories surrounding these fancy auto-completes.

This comment is confusing to me on so many levels. What’s with the tangent(?) about a math test your algebra teacher could have generated? Did you bring up an illiterate teacher (extreme outlier) as evidence that the general population has low comfort with written English, or…? (I’m going to resist getting into the rest of the tangent, but it’s really impressively densely perplexing.)

(edit: I’m not going to resist)

> If he could have written (in 2009) "give 20 question test for week 1 algebra II student with answer guide"

Is the “could” here just about AI not existing back then, or does “could not interact with the written language” imply that he could not have written this prompt? Why would he need the output, given that most of it is math? (If we assume he can speech-to-text the prompt, why can’t he do the same for other writing?) If the level of writing of “Write equation of a line in slope-intercept form with slope 3 and y-intercept −2” is the challenge, is he able to read it? What if the output is wrong – who’s going to verify it? Are you presenting this as a good thing? How did/would he grade handwritten written-answer questions?

People won’t get better if they don’t use it. These LLMs are a crutch that will not just not people from getting better, but any skills they do have will atrophy without use.

How can anyone even write a proper prompt, or understand if the answer is correct, without being literate? I’ve been noticing on YouTube where people are delivering scripts that are clearly AI written. It makes me question if they are able to read something, understand it, and put it in their own words. This seems like a fundamental skill any adult should have, especially if they are trying to make a living giving advice to others, as was the case with these videos.

I’d also think literacy should be a basic requirement for a teacher, regardless of the subject. If we don’t hold that standard for our teachers, how can we expect it of our students? Continuously lowering standards is not helping anyone in the long run. It hurts the individual and society as a whole.

What does this have to do with emails. The comment you replied to didn't say to purge all LLMs so I see basically no connection.

As a non native English speaker, letting an LLM translate my emails is actually a sweet trap that’s all too easy to fall into. But Gemini in Gmail DOESN’T SUPPORT non English input. What’s the point of this thing?

>As a non native English speaker, letting an LLM translate my emails is actually a sweet trap that’s all too easy to fall into.

There's a big difference between running your input through DeepL and asking [ChatGPT] to rewrite it. I don't know how we got to the latter being the go-to

DeepL's limit is now only 500 words. And I already have a whole bunch of subscriptions. This will become a habit. That’s why DeepL is feeling the pressure and [laying off](https://www.bloomberg.com/news/articles/2026-05-07/google-tr...) employees.

If i write a bad email because i'm frustrated to some company or whatever and want them to change their behavour, i think a llm can write an email, which triggers these people a lot more than my 'polite' way of convincing others.

Reply on news.ycombinator.com