[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/g/ - Technology


Thread archived.
You cannot reply anymore.


[Advertise on 4chan]


File: images (1).jpg (29 KB, 511x391)
29 KB JPG
>PM asks me to verify action plan sent out by sales people
>reading this mail, I can just feel it in my bones that it's A.I. slop. Sure it doesn't have the million dashes, but still it uses gay language "avoiding unnecessary....on the broader...." and a lot of wordy nonsense and over use of bullet points

Makes me fucking rage how there are people getting paid to spam LLM slop and others like me then have to detect any mission critical mistakes in the "plan"

Fucking shit I hate this so fucking much

>in b4 quit your job and don't complain

Yeah fuck you too cock sucker.
>>
>>109036141
This is a humiliatng way to live and I would either quit and move to Portugal or kill myself in your situation.
>>
>>109036147
>quit and move to Portugal
What would I do in Portugal? Life is not much cheaper there and there are almost no opportunities to make solid money
>>
>>109036141
You think that's bad?
I'm an IT-consultant for 15 years.
I always wrote and spoke polite, on the point and made small, actionable bullet points for my always busy clients. In difficult meeting I always made sure everyone's opinion is heard and relevant parts praised even if we move into a different direction, as to have everyone on broad. The way I work was highly appreciated.
Now people think I just copy paste everything from chatGPT, even if my content is based on deeper understanding of my clients needs rather than hallucinations.

I actually asked an AI how to cope with this and it told me to make more mistakes to appear more human. Kek.
>>
>>109036190
OP here
>business critical service outage
>few experts going back and forth on how to solve in group chat
>see director typing in group
>stops typing
>sends me directly instead
>wall of text from chat gpt with its trademark hyper confident and condescending approach to the problem "oh this is a classic case of xyz, the solution is abc"
>totally wrong not even close

What the fuck are these morons thinking
>>
>>109036141
Ask PM is I am supposed to compare this to an existing plan or what is the point of this writeup (since when are business people writing the cutover activities?)

Tells me I need to compare it to cutover plan on teams

If there is a fucking plan why aren't we following it

Let me now dig through dozens of unorganized teams shit
>>
Maybe I will reply again next week. Tired of this shyte
>>
>>109036141
Your email also reads like fuck off AI slop.
>>
>>109036190
>Now people think I just copy paste everything from chatGPT
This.
There was no point in me improving my English I guess. There's no point in putting effort into anything anymore since people just immediately look for ways to discredit it.
It's a nice compliment for someone to assume you're AI, but it's also annoying.
>>
>>109036167
I have significant passive income in USD.
>>
>>109036307
That's the point
>>
>>109036348
Was this all just an excuse to be able to share that information
>>
>>109036141
any time i get a hint of text being AI generated, i immediately stop reading.
>>
A cuck just sent a farewell email using LLM (last day on his job)

Fucking moron
>>
Sometimes I write my emails with white text so you have to highlight it to see what I said.
For the same reason people used to write with invisible ink.
>>
>>109036559
Do you sometimes sneak in some insults
>>
>>109036190
Around 2022 I was often accused of using a chatbot to type because my typing style is 'too structured'. But I've never been accused of being a bot after that, I assumed it was because people realized these clankers can't write anything meaningful, while my messages are usually pretty deep.
>>
>>109036216
My boss does this all the time lmao
One time he told me to point claude at a problem related to a bullshit web portal virtually nobody uses that changes every 5 seconds and stopped supporting basic features according to even the official docs. I asked him how am I supposed to point claude at it to be polite but holy shit I wanted to rip his head off. He had no answer of course.
>>
>>109036552

Every sentence was a paragraph literally every single one
>>
>>109036141
I’m a anti-AI luddite, write all the mails myself.
I still end up with a lot of bullet points, they just allow you to make the messages so concise it’s hard not to when everyone is constantly bitching about mails being too long

am i cooked
>>
>>109036592
>everyone is constantly bitching about mails being too long


Let them bitch
>>
>>109036141
Using an LLM to draft an email isn’t the problem. Sending unread, generic slop is the problem.

A calculator doesn’t make someone good at math, and an LLM doesn’t make someone good at communication. It’s a drafting tool. If you review it, correct it, add specifics, and make sure it actually says what you mean, there’s nothing wrong with using it.

What you’re mad at is not “LLM emails.” You’re mad at people outsourcing judgment.
>>
>>109036592
You’re not cooked for using bullet points. Bullet points are fine.

But refusing to use an LLM on principle is just deliberately handicapping yourself. It’s like insisting on sewing by hand when there’s a perfectly good sewing machine right next to you.

The problem isn’t using the tool. The problem is sending whatever it spits out without reading, editing, or adding your own judgment. If you use it to draft faster and then actually revise the email, that’s just basic efficiency.
>>
>>109036607
The problem is that the LLM never produces a good message skeleton, ever. It's always total garbage. Calculator is more akin to grammar checking tools. And indeed, using LLMs for grammar checking wouldn't be a problem. But another problem is you can't use them for translation or correction because they often refuse to do their job if there are badong doubleplusungood words in your document. Even something as simple as 'blacklist' can #trigger them sometimes (depends on rest of context), it's crazy.
>>
>>109036619
My idea of handicapping oneself isn't exactly in line with "making myself faster, more efficient, and more correct" but hey, you do you aitard.
>>
>>109036619
>But refusing to use an LLM on principle is just deliberately handicapping yourself. It’s like insisting on sewing by hand when there’s a perfectly good sewing machine right next to you.

the thought that one hopes to communicate via email lives in one's head. outlook has spell check and grammar check built in. what are we left with -- using an LLM to fluff up an email with filler? emails should be terse, and so LLMs are antithetical to good emails.
>>
>>109036190
your post does read a little like an LLM... but the issue is that the LLM is copying YOU. when they're asked to do professional writing, they output something like what a professional might write, with lots of corpo-speak, marketing-speak and whatever it's called when you write something negative in a nearly condescendingly nice way.
>>
>>109036190
This is a actually the killer, where correctly worded, grammatically sane full sentences are "lol bro dats ai slop" despite having none of the trademarks like "triplets", "it's not x it's y", em dashes, and flowery redditspeak paragraphs and giving a succinct, actual analysis of the task at hand
>>
>>109036686
It has none of the "orthographic" hallmarks, and is actually making a point instead of waffling, you're again falling into the trap of "he used proper English" to flag him
>>
>>109036619
>Not spewing out a wall of redditor waffle is inefficient
>>
>>109036619
Just put the effort in and learn, i did, now writing mails by hand is quicker than tard wrangling an AI and correcting it’s mistakes.
We have ChatGPT at work and i tried using it, but even the simplest tasks i gave it required a near-total rework, the results couldn’t be trusted.
I need to react within 30 mins, so i just need to suffer&learn until i just know the solution instantly. Tard wrangling an AI and double-checking it’s output means i’m already too late.
>>
File: Fm_4ur-XEAM9pby.jpg (57 KB, 1200x675)
57 KB JPG
>>109036141
You're supposed to be asking AI to summarize the email for you
>>
>>109036141
Our HR department has not published a single word that's not generated by LLMs. All announcements, all wiki pages, all correspondence, everything. It is so tiresome. Why are they even hired, they are essentially interfaces for chatgpt.
>>
>>109036897
I use both bullet points and em dashes oh god oh fuck
>>
>>109036141
it's just NPC things
>>
>>109036216
once I had a question to one of managers, he just asked
>what's your question
and when I told it he typed it into chatgpt
completely deranged behaviour
>>
>>109036619
Sewing is an incredibly rote and repetitive task no matter how efficient you are at it so automating most of the process is welcome. A sewing machine can make 10-20 stitches per second. Those are astronomical gains for little effort and humans cannot keep up with that. Rejecting an AI when it comes to writing 10 sentences is not a comparable scenario. You're being extremely dramatizing and hyperbolic in ways that aren't grounded in the reality of what we're discussing which is why reading your post feels like AI.

Formatting a few sentences of speech and breaking up the core ideas in separate bodies is something the average person can easily become very efficient at doing. Proompting adds extra complexity and multiple steps of review afterwards, and still requires you to input what you want to say to the llm in the first place. It is not comparable to rejecting a calculator.
>>
Ayo. Bro. I work with a guy that sends llm-made slack messages ... that shit is borderline psycho, like not even the minimum effort for a one liner
>>
>>109036619
>It’s like insisting on sewing by hand when there’s a perfectly good sewing machine right next to you.

Stupid analogy because it doesn't address the point. People should wear clothes stitched by a sewing machine because they are good. Should people read emails written by an AI? Obviously not when you can use an AI to summarize the email for you. So we basically wasted electricity and blew up computing prices to transform relevant information into something humans shouldn't entertain and then transform it back into relevant information.
>>
>>109037189
>tell an AI what you want to say
>it turns it into some crap nobody wants to read
>recipient asks ai to summarise the crap into basically what was actually said
>>
>>109036686
almost like LLM "learns" from humans or something wow!
>>
>>109037277
I've gotten so many replies from people repeating my own point as if it's something i hadn't considered that it's actually starting to concern me.
>>
>>109037098
Em dashes as bullet points or when appropriate is reasonable, trust me you'll know when you see the AI overuse
>>
>>109036141
I personally love weaponizing LLMs and making them spew out pages worth of text that I send to people that waste my time. You asked me the same question I've answered to you previously half a dozen times. Here, have this 10 page long response that might or might not have the solution.
>>
>>109036141
I'd use an LLM to summarize slopmails
>>
>>109036660
This. The comparison is wrong because something like a calculator and LLMs are completely different in philosophy.

A calculator, or anything of the sort, is supposed to give you a precise answer to a problem with well defined semantics that is objectively correct. That you can blindly trust. Doing the calculation yourself afterwards to "check" whether the result is correct would defeat the whole point of calculators.

By contrast, LLMs generate content (images, text, etc.) which almost never objectively answers the question: whether the result is correct or not is a matter of subjective interpretation, and this is something that, given the current state, cannot be offloaded to a machine.
>>
>>109037090
How is this a bad thing again? I would love to be able to cut out the AI encode:decode layers and just send emails like this but the 5% of people who demand that you talk like a good little robot get so pissed off if you don't that the other 95% of us can make good use of this.
I feel bad for the luddites that now have to wade through the sea of slop but not bad enough to give up a tool that lets me write blunt and concise emails without dealing with the shitty 5%.
>>
>>109037361
I will never trust am emdash supposedly written by a real human
People do not use that in normal speech
>>
I use a locally run qwen 3 model with a system prompt based on my writing style to write 90% ofy emails and reports
>>
>>109037609
you can't enter an emdash using a typical US layout keyboard, and i say this as someone not in the US. any keyboard intended for english-usage i know of is us the US layout or very similar to it.
in fact, i don't think i even knew was an "emdash" even was before the advent of LLMs over-using them.
>>
>>109037609
It depends on the context, some typesettings do "do" em dashes and there is a use for them, but when you see one just thrown in there with reckless abandon it's definitely an LLM, though that's not even the biggest giveaway
>>
>>109036607
>Using an LLM to draft an email isn’t the problem. Sending unread, generic slop is the problem.

>A calculator doesn’t make someone good at math, and an LLM doesn’t make someone good at communication. It’s a drafting tool.

If you couldn't trust the accuracy of a calculator and had to check its output by hand, who would bother using them except morons who don't bother checking because they don't know how?
>>
>>109036375
No I'm not even the original anon
>>
>>109037609
I've been using em dashes for many years, long before the rise of LLMs, I've always liked their versatility.
>>
File: this-is.png (60 KB, 418x360)
60 KB PNG
>>109038089
>in fact, i don't think i even knew was an "emdash" even was before the advent of LLMs over-using them.
Emdashes are used heavily in academic writing, which is one of the first places LLMs scrubbed for data.
When I did my masters I saw all sorts of weird formatting and punctuation that just does not get used in casual writing as is 97% of the internet.
So if you never did academic studies or read white papers you very likely never encountered one.

>>109038304
Calculators are solved because math is an entirely logical system of information.
If the retard normies expect every subject known to man to also be solved because muh chatbot said so, then it's an uphill battle to convince them otherwise.
>who would bother using them except morons who don't bother checking because they don't know how?
Except there's plenty of morons smashing buttons on a calculator who don't have any clue what the fuck they're doing, any many people who don't correctly understand what they're asking the calculator to do (i.e. using the wrong syntax). Now we just see those same idiots en masse because more people own an online device than a (proper) calculator.
>>
>>109036619
>The problem isn’t using the tool. The problem is sending whatever it spits out without reading, editing, or adding your own judgment. If you use it to draft faster and then actually revise the email, that’s just basic efficiency.
There is no way using an LLM actually speeds up your work UNLESS you are sloppy and just accept some of the shit it puts out. People just don't want to admit it.
All of the supposed benefits of AI are actually "benefits" of just not giving a shit anymore.
>>
>>109038671
All these arguments against llm are fucking retarded, it's the most basic of task delegation, and these can only be espoused by someone who has no experience in the real world.
Good luck arguing that having subordinates doesn't save you time. I guess you know better than everyone who has ever had a secretary or scribe write for them.
>>
>>109036141
I only write emails with a llm if it's to someone I hate.
>>
>>109036592
>am i cooked
Why are you anti AI? You already talk like a generic bot, carbon copy zoomoid.
>>
>>109037609
we're kind of fucked if LLMs stop using em-dashes and stop sounding like redditors and HR karens
>>
>>109038606
>Emdashes are used heavily in academic writing
No, they abso fucking lutely aren't.
t. PhD comp sci.
>>
>>109039542
In the sense that it would take 4x as long to know we're wasting our time, yes. Which is the real problem with AI. Can't write a single word worth reading, but can't find out if there is some hidden useful nugget in the bullshit without consuming it or getting to a point where it's clear it's AI (= worthless guaranteed)
>>
As somebody that has been accused of writing like an LLM, fuck LLMs. They stole my identity from me. And furthermore, let me list my thoughts in a series of nested bullet points and diagrams.
>>
>>109039542
You could always train a LORA to write the same way you write but for now anyone doing that is smart enough that even their slop output might be useful.
>>
>>109036141
For business, whatever, cunts will be lazy shits. Ignore and move on.

I have family members texting me generated shit. Just bring on the apocalypse.
>>
>>109036348
You wouldn't if you were
>in your situation
you retard
>>
>>109038089
I have a custom layout which is my default that allows me to type en-dashes with AltGr+Hyphen and em-dashes with AltGr+Shift+Hyphen, among tons of other things.

I always take the precautions nowadays to add a footnote with the first emdash in a report/presentation that everything in the work was done by a real human bean and I'm a prim and proper anti-sloppa retard.
>>
>>109043526
Not true, only a special layout has those options. In a normal layout, altgr - and shift+altgr+- produce no output. Take your schizo meds.
>>
>>109037090
diplomatic autoencoder
fucking kek
>>
>>109036348
Usd is worth less than euro just fyi.
"American passport waving" is a total Hollywood invention too. If anything, Americans tend to be silently unwanted or banned from places, all over the world.


>>109036141
PMs are the first on chopping block of jobs aren't they? Not really management, not really contributors.>>109036141
>>
>>109043921
What part of
>I have a custom layout which is my default
did you have trouble reading you absolute fucking moron?
>>
>>109043994
So you agree that you trannified your layout for the purpose of plausible deniability when you post slop.
>>
>>109036348
thanks for informing us
my associates will be with you shortly
>>
>>109038606
>>109043526
sure, i'm sure i've seen them before, and probably even had them auto-inserted, what i mean specifically is that i didn't know their name before, their sudden over-prevalence has made people talk about them where they were relatively invisible before.
>>
File: 1780884016696309.jpg (42 KB, 800x450)
42 KB JPG
I just don't want to work anymore.
>>
>>109044312
I just want to get funding for my startup so I can work on that instead of spending 40 hours a week at a day job to be allowed to moonlight on my startup getting 0 sleep because the clients are going to leave if I can't deliver on time but somehow making ai for content moderation (no prototype) gets funded before something real like I have.
>>
>>109036921
This, less is more, all Ai can produce is essay slop that nobody has time for.
>>
>>109036141
It's merely a tool. Choose to be happy instead.
>>
File: file.png (105 KB, 938x446)
105 KB PNG
>>109044010
Your brain has melted under culture war stupidity.
I've had the layout since pre-slop era anon.

It also has "I sh", "iggy", " dig", and "gy." bound to AltGr with four separate keys serially (It was made in MKLC, can only have 4code points per key), that's how old it is.
>>
>>109044312
Same. Why bother ripping my ass getting better at something if that gets automated next year.
>>
>>109036141
Just decipher it to human language with another AI.
human->AI --> corporate slop text --> AI->human
A new kind of protocol, the future you chose.
>>
>>109044852
OK slopper whatever you slop-say.
>>
>>109045025
You got me. I was an LLM all along.
>>
>>109036348

why portugal? I'm portuguese and let me tell you that you'll struggle to integrate. portugal is nothing more than a theme park for rich foreigners now. portuguese culture is completely dead. it's as souless as dubai. you'll be surrounded by both """expat""" americans and indians who don't speak a word of portuguese.
>>
You just reply saying I will not read machine made garbage

If you couldn't be bothered to write it why would anyone be bothered to read it

and fuck your emojis in the emails you cunts
>>
>>109036619
Why are people sincerely replying to an obvious AI comment?
>>
>>109046218
I did that with a customer service AI email bot which still replied to every email, and eventually ended up getting confused and began answering every email twice until the business owner finally showed up pissed at all the tokens I was burning
>>
>>109036141
I don't care. My job is a complete waste of time. Enjoy the AI slop proposals and emails as it's not worth investing my time into them.
>>
>>109036141
>no cool down period. No hard feelings.
slop
>>
>>109045394
I wouldn't actually leave America long term. For all of its problems I don't want to jump ship because things are crazy. I could see myself living overseas for a few years however.
>>
>>109036141
Blame boomers who made culture of mails required to be full of shitty formality fillers instead of right to the point
>>
>>109047244
property is so insanely fucked in Portugal. I beg you, do not fuck the housing and renting market more bro.



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.