Don't paste the AI, please
dontpastetheai.com
dontpastetheai.com
Write as yourself. You’re being paid for your expertise and insights. Communicate them directly to us. Copying and pasting Claude responses into Slack or an email directly shifts the burden of comprehension and understanding to everyone else, and worse, risks skipping that step for yourself. Even if you’re fundamentally using Claude to gather your thoughts or help you prepare a response, you need to be writing it yourself, in your own voice. Not having Claude ape your voice, or “make it sound less like AI”. You, directly. Doing this will further reinforce your own understanding of the state of things, the same way teaching someone is the best way to learn. As a guideline: for Slack and email comms, this should be near-universally written as you. For deliverables that are longer form and follow a template like proposals, roadmap/discovery work, etc., use of agents is expected but, see Principle #2. (Own the Output.)
It's very likely that proper punctuation and capital letters help the reader, then it doesn't matter than not using them is faster to type.
I'm sure we've all gotten that big stream of consciousness email that's a 15 line blob of lower case text, with no punctuation or line breaks.
I don't mean every ephemeral message needs to be perfect, of course. Just, you know, readable.
You can also teach those people about text expansion software. macOS and iOS have it natively, though if I’m not mistaken it doesn’t work properly in non-native apps such as Slack. Alternatives like Alfred should work anywhere. That way they can keep using their abbreviations to type faster, but others won’t have to deal with them.
> It's very likely that proper punctuation and capital letters help the reader
Can confirm. Not using proper punctuation and capital letters makes reading less fluid. I’ve stopped caring about posts and comments that do that—if you don’t care enough about what you wrote to do the most basic of courtesies (and sometimes even fighting the auto-capitalisation that’s done for you) then why should anyone else expend the effort?
You asked a human for their advice- based on their judgment. If they don’t take the time to validate the answer provided by ai, they literally have no judgment. A transparent proxy has zero value.
It should be the other way around-- you give the AI instructions and it does stuff. Any time the AI is giving you instructions and you do stuff, that should be a big red flag.
I can understand that there are situations where you just can't type out the answer and those exceptions are of course fine, but if you post a voice message or god forbid a video message in a group chat you are forcing everyone to take the time and sit through it.
To the semi-literate a few paragraphs of text represent a significant investment of their time.
I suspect we're both right, just on different occasions.
I hate genAI deeply, for many reasons, but one of them is the way it bastardizes and ruins the written word.
It writes incoherent, worthless prose, taking hundreds of words where two sentences would do.
The signal-to-noise ratio in genAI writing is just wretched.
Yes, I can read the thousand-word ball of muck you pasted from Claude in about thirty seconds, but it could have been one second had you used your mind instead of committing a worthless act that cost me time and gained both of us nothing.
Either you haven't used GenAI in a few years, you've only used bad ones, you're really bad at using them, or all of the above.
> I hate genAI deeply, for many reasons
Probably explains the above issues. In my experience people who don't want things to be good:
a) Always find them to be bad
b) Are always bad at using them
I use Claude every day, by corporate mandate, generally with Opus 4.8 or 5, but sometimes Sonnet 5 (per corporate demand to both use LLMs heavily and simultaneously keep the bills within a budget).
I hate doing it, but I get results my coworkers, PMs, and end users are happy with.
> Probably explains the above issues. In my experience people who don't want things to be good...
I admit regularly that LLMs do amazing things. They're astonishingly good at breaking security, for instance.
That doesn't mean I like them.
And no, they really are not good at writing.
When you feel the impulse to be defensive, that's often a sign that asking questions will be more worthwhile than leaping into the defensive posture.
I don't mean to be defensive, just to point out that most of the statements you made are provably and obviously untrue... And given that you admit bias that's the most likely explanation.
I hadn't considered that you have experience, but only with one that is legitimately bad at this particular thing because someone forced it onto you. Fair enough.
We don't all have to like the same things.
So yeah, maybe every single person in my company sucks at Claude, or maybe Claude just actually sucks at writing and you don't know what bad writing looks like
So if the goal is "writing" and they are using Claude, then they do in fact suck at it.
Its like using a butter knife instead of a screwdriver. It might work if you waste enough effort, but will never do the same quality of work a screwdriver can.
People are just afraid of billionaires becoming trillionaires and not having food and shelter
So their output from an llm has the halo/ikea/endowment effect and they want to show off what they “wrote.”
Writing is actually super easy. Editing and writing well is the hard part. It’s like co-workers who wrote a terrible book being excited just because they wrote a book.
Too bad that’s not more universal.
It's so normalized there that I wonder whether this is a particularly Spanish habit. Maybe there are a bunch of countries where people routinely leave each other voice recordings on WhatsApp/Messenger/etc, and I just haven't been to them yet.
1) long voice memos over succinctly typed notes
2) not forming lines/queues and just mobbing the train/shop/etc
3) habitually late or not on time
Also even queueing can have different norms. Here people rush to a newly opened supermarket checkout like their life depended on it. I feel that should be a lot less sociably acceptable than it is, especially for those with overfilled carts that will add huge delays for anyone behind them.
Fortunately my colleagues were very understanding, and I reverted back to text once I could type reasonably well and with minimal pain again.
Also, circular sawblades are terrifying.
They were putting every question I asked into AI and pasting it in slack
Very sad
2. Own the Output. Everyone is encouraged to use these tools to accelerate the pace of their work and to take the toil out of their daily output. But that doesn’t mean you simply defer to the model and skip your own careful review step. Part of the paradox of these models is that they’re wordy and it can feel exhausting to read through everything they’re writing and confirm its validity. But that’s an enormous part of the job: you’re the human-in-the-loop to make sure what it’s saying is correct, and that you’ve reviewed it before you’ve shared it with the rest of the class. Read all of the words before they reach the rest of your team or a client, and make sure you stand behind having your name attached to them. In situations where a full review isn’t possible for expediency, disclose it directly! “Hey, I haven’t had a chance to review, this is pure Claude, but here’s what it says.” This isn’t a get out of jail free card, but it can be used occasionally when the situation dictates. Most of the time, you need to review and adjust the output before other people see it.
3. Be the expert. Have an opinion. This is a big one. Do not offload your role as expert to the model. The model is an assistant. You need to internalize the work being asked of you, and apply your education, instincts, and specific expertise to form your own opinion. You can use models to work to adversarially challenge things, but part of being an expert is knowing when to discard irrelevant feedback. Use this skill daily. Don’t let it atrophy.
4. Don’t Slop-Bomb. This is an amalgamation of all of the above, but again, don’t just spew large volumes of model slop into comms channels. Bring your own perspective, internalize the problem in front of you. And be judicious about what you share, and how wordy these things are. Put another way: edit for clarity and concision. It’s too easy to generate busy-work-seeming reports and outputs, and that burden shift to everyone else is multiplicative. Use fewer words, distill things into insights (backed with longer form deep dives), but make sure you’re not just making the rest of the team read a bunch of useless pablum. I don’t want to have to make my model read your model’s output to glean what I’m supposed to out of it. Do that for your coworkers and your clients. Think in terms of information hierarchy.
For purposes of training the AI (enablement firm)?
> Write as yourself.
Unless it's your first day on HN (it's not 2008 account), most people use paragraphs and sentences longer than 2-5 words. Something like "You, directly." in the middle of all that has a certain vibe to it.
> Copying and pasting Claude responses... directly shifts the burden of comprehension
Most definitely.
The problem is we live in a world where people try to pass off copy pastes as their own all the time, many even smart enough to remove the em-dashes now. So I don't think that's going away soon. Your first sentence seems human, the rest doesn't match the tone at all.
Isn't that the very thing you just claimed people don't do?
For simple, impactful sentences, I often use short ones. They work better.
Perhaps that speaks more to your education and that of "most poeple on hn"
https://www.anthropic.com/learn/claude-for-you
This framework provides a more abstract way of thinking when working with AI systems. Especially valuable to me is the pattern of the 4 D's: Delegation, Description, Discernment, and Diligence, which I have now internalized whenever I use AI systems. It helps me understand my own and others' (mis)use of AI systems a bit better.
Like is your insight that copying and pasting something is not understanding it? Do people seriously not known this?
There is no need to Socrates when writing policy docs. Maybe this is why people ignore reading your actual voice messages that you feel a need to also justify this basic request. Sometimes the human voice thinks too highly of itself also.
And maybe GP doesn't want that.
What you wrote just seems common sense to me, and I'm hopeful it will emerge as such given enough time.
The effect of this over time is cumulative. Cervantes said (in translation), "Show me who you spend time with, and I'll show you who you are." I realized as a teenager that the books I was reading and people I was listening to were shaping my ideologies in ways that were outside my control. It's prohibitively difficult to pause when presented with every idea and analyze it to see whether it makes sense. If you're cautious you'll catch some ideas and analyze them, but a lot of the time, you won't even consciously recognize that you've been presented with a new idea, and you'll simply integrate it unconsciously. Time passes and it becomes difficult to remember where our ideas came from or why we thought they were true.
This was a problem before AI: if you surrounded yourself with people who thought poorly, you thought poorly. Now with AI, millions of people have let a "thinker" into their lives which has massive information and no wisdom.
Frankly, when I find out someone uses AI in any way, I lose respect for them. The best case scenario is they use it similarly to the post I'm responding to, filtering the ideas they get from AI through their own intelligence. But that doesn't really work==original thought it harder than that. And that's the best case scenario--most AI users don't have anywhere near the discipline for that, and just regurgitate what AI tells them.
Rewording it doesn't help--it's still not your idea, from your experience and logic. And once you use AI at all, it's very hard not to be just rewording.
I hate spell checkers, I disable them and I do manually double-check words that I'm unsure about. But I'm under no illusions that most people are going to see automatic spell-checking as useful and time-saving, and I'm not going to waste my breath trying to convince anyone otherwise.
Where they disagree is if you add value on the knowledge that makes the prompt possible or add value in rewriting AI to pretend you wrote it.
If an AI answers a question perfectly, I'll go with it. But if my boss fires me for no longer being needed because of the AI, good luck to him and his new idiot employee.
Asking questions is part of the job. Answering questions is part of the job. Collaborating with other people is the most important part of the job.
Some junior technical folks early in their career might bristle at this: "my job is to write this code I'm writing, and anything else is a distraction", or something along those lines. It's hard to internalize that your job is actually making the company successful, not yourself, and that means making other people successful.
If you can't answer a human question from a colleague, as a human, then the natural conclusion is that you are unable or unwilling to do the job. Like, great, you can write (or generate) code, what about the other 90% of the job?
That's for their manager to judge. You can maybe share your opinion on the matter, and let them decide. Until then, collaboration is more important than your personal IC work, and usually more effective than 1 person working heads-down and another struggling.
Newer devs invariably discover that the "write the code" part of the job is just a small part of being an engineer, and less value-added than collaboration. Big successes are achieved when you make others successful, rather than isolating yourself and focusing only on your own work and your own success.
Also, if the response of AI is sufficient, what is the added value of you spending more time on it?
The only reason someone needs to write this is because leadership is not smart enough to solve the problems it showcases, in a way that is respectful to how the employees with years of experience learned their craft.
They have a tried and tested thought pipeline that makes them capable of what you hired them for, and you try to introduce a magic eightball to that and then force them to research that whatever the statement that shows up on the ball is true before they relay it to someone else. It’s unnatural nonsense.
Ingenuity comes from that spark in your brain that happens when you really dig into something because you’re interested in it. A machine will never be capable of that. AI didn’t invent a single thing, it’s simply finding stuff in all the data the labs stole that is already there and humans overlooked.
Now you force people to start working with whatever the magic word generator put out and go from there. They have zero incentive for that, because those thoughts come out of a black box and are not their own.
I find it ethically troublesome you somehow see the need to force behavioral change down the employees’ throats because company leadership is collectively failing to address the core of why people paste AI into everything.
They have nothing but disrespect for the technology, because it makes their job harder and them as a person dumber and less satisfied with their work.
Hire an external company to do interviews with each employee about their use of AI in your company and how the way your company is using those tools annoys them. You’d be surprised what you find out, when people have no repercussions to fear and anonymity.
There is huge resistance if you see it necessary to write a document for the whole company that forces behavioural changes.
Sure, you can hit the dog until it does what you want, but if you do that long enough, you get your head bitten off.
You need to find out why people are doing all these things, and then fix them. No extra paperwork is needed.
Seriously so many people need this tattooed on their forehead so they’re reminded when they look in the mirror. I am so sick of it.
The only thing worse is people who literally just copy and paste a link to their ChatGPT session. Wild behavior.
It was wild, as it had never even occurred to me to just copy/paste out of an LLM while having a convo with someone else.
What is worse is that they don't even bother to provide the correct context to the model, or even use the same model to answer your questions. So, for example, they submit a PR using an LLM, and when you ask them to justify an engineering decision, another LLM responds with a completely nonsensical reply because it doesn't have the same context as the first one.
What if the response was perfect and sounded exactly like your colleague. Would you care if it was AI generated then?
I think there could be a world where a colleague could just ask its (finely tuned) AI tool of choice to respond for it and colleagues would hardly notice as the output would be so good!
I can’t help but think that that world would be so much more productive as we would spend less time in slack answering messages and more time doing more important work.
However there are two very important things that need to be in place for this to happen: - colleagues have to have their AI tool know all of the their context on their projects….etc and and in-depth understanding of how they write - colleagues must review (and tweak if necessary) the output
My guess is that the problems you are seeing is that your colleagues are failing on one or both of these points
It used to be fairly obvious (using em dashes, etc) but now AI writing, with the correct guidance, is very much the same as regular conversational writing.
And if you really want to record all your work and project context somewhere, just let people ... look at that directly? Or summarize and query it with their own LLM if they want to?
What's the equivalent of "Google it yourself" these days? When you ask something of somebody, the expectation of the other side to drop everything and do work for you isn't always appropriate/balanced/appreciated.
The underlying problem here is that communication is hard; especially if it's not face to face. People can't read your mind or might not get the full context of your question.
Back in the day I learned the hard way that sending a long reply to an email just ensures the other side won't read it. You can't assume that you dealt with their request effectively that way. And that was when there was a lot of email to deal with. Lots of people demanding all sorts of things via email and creating more work for each other by sending more email.
I learned to 1) never reply instantly to low priority requests, 2) be very brief and clear, 3) that talking to people is almost always the better choice when you get tempted to write a long reply.
If you don't like the answers you are getting, maybe reflect on what you are asking, how reasonable that request really was, and how you communicated it. Maybe your question was entirely reasonable and the answer really inappropriately bad. If so, give feedback or talk to the person. But maybe the other side didn't get you. Or maybe they were busy and you are interrupting them. There can be all sorts of reasons why you aren't getting what you expected.
It all depends on the context of course. If you are delegating something to somebody that reports to you, you can just tell them off. But if the other side is a colleague, your boss, or a friend you have to be a bit more considerate.
Do you actually say that to people?
Edit: i’ve hit the comment limit per given time, but all I have to say is “what have you found so far?” And “Google it yourself” are radically different responses man. The former is far more reasonable, it shows you are willing to be helpful and is a productive question. The other is just rude and implies your time is too valuable as well as assumes they’ve done nothing to tackle the problem. If you actually say “Google it yourself” when people ask you a question that would be absolutely bizarre behavior.
A frequent first question when a subordinate asks me for help:
“What have you found so far, searching for answers?”
Thanks for reminding me of lmgtfy. We used to say that all the time 20 years ago, and it was way politer than how the people older than us spoke. It took some getting used to, but I really miss the hilariously mean bluntness of the IT/tech world.
In retrospect, I think it was just a bunch of people who grew up breathing in leaded gas fumes, with undiagnosed mental disorders. But I miss it just the same.
That was never an acceptable thing to do outside of light hearted joking. You’d do that to a colleague or someone asking for help in earnest?
It’s a deliberate, rude act that is aggressively proclaiming “how dare you waste my time.“ God help you if you ever ask a question you could’ve looked up yourself, because of course you never do that right? You always make sure to engage in rigorous scholarly research before daring to bother somebody.
I almost just said it was like working with The Sopranos or Dr House, but now that I think about it maybe that's because I was in Jersey?
How on earth are you using an absurd work of fiction as a good thing and then justifying treating people like shit at work? People don’t want to be cursed and yelled at and personally insulted at work and they shouldn’t have to deal with that.
As I think about it, I guess I really just miss people being more authentic. There was a lot more understanding and forgiveness around unmedicated emotional outbursts. Modern work culture is more polite on the surface, but it feels like everyone is a cut throat politician manipulating their way to the top with a dumb fake smile. I would much rather work with someone who is upfront about their plans and how they're feeling, than someone who is friendly to my face while talking shit behind my back.
I do kinda feel bad for the kids that are entering the workforce, expected to act like emotionless robots, while spending most of their time talking to AI. There is an underlying sense of fear that they have to mask themselves to not be fired.
But yes, it is an overall improvement. I did have one truly abusive boss that was able to hide unchecked within that culture, and I would not wish that experience on anyone. The rest were just normal people that were emotional and funny.
I understand what you’re saying, it is fun to kind of have a little bit more of an honest, “in the trenches” culture. I miss that from time to time from my work on film sets. But I can also say that the lows professionally were far lower as a result, as those blurry boundaries can cause much more friction and even harm people. So I just have a kind of a strong opinion on that
If your question shows that you haven't done basic research, because your question would be answered by the most basic of research (like reading that entry in the manual), then you're wasting my time with your laziness. That might be regarded as deserving a bit of rudeness in the answer. (The exception is with someone who is so new that they don't even realize that there is a manual, but that should be very rare.)
I mean, hopefully I've grown up a bit, and grown more patient a bit, but I still understand why people reply that way.
You added this to the conversation. That was not the original discussion.
Also you and I both have absolutely asked people questions we could have looked up but we knew they likely knew offhand. I don’t understand why some of you think your time is so precious that’s every single question ever asked of you must meet some minimum, arbitrary bar you’ve set internally before you deign to come down from the mountain and bestow knowledge upon your friends and family.
For the company it would probably pay off better (over time) to let the people who regularly get stuck after trying nothing spend some energy figuring out than what they get in return for shipping them to conferences.
Additionally, when it’s not your domain knowledge, research is more difficult and time consuming. It can be difficult to even know where to begin. Someone new to media production, for instance, wouldn’t know the major forums to check off hand. Someone had to tell me about dvxuser, reduser, cinema5d, etc.
We learn through a combination of our own effort efforts and the help of others. Otherwise why would we even have school?
Love it, thank you!
A company isn't just a bunch of people isolating themselves and doing their best to not talk to each other.
Yes they are. "RTFM" may be an angsty insult thrown around in some toxic open source projects, but it's rude and inappropriate for an actual working environment where you're supposed to (and being paid to) help others. Most time spent at a company at a normal job is collaborating with others.
Your time is not so valuable that you cannot use it to make other people and the company successful, and when you're on the clock, that time belongs to the company anyways, who would likely prefer you do what's best for the company, rather than yourself.
That said, "what have you tried so far" is a good question to ask, too, to coach your colleagues on how to find info, and to get context on what would be retreading their steps.
The question is whether one employee's time is more valuable than another employee's time. One employee answering another employee's question expends two people's time. If the answer is in the book or on the first page of google, it is wasting one of those employee's time, and that is the boss's money. If the employee asking the question makes ten times as much as the employee answering the question, and it saves him 10% of his time over flipping through the manual, it's a wash. If both employees make the same amount, or the other person even makes less, it is obviously a waste.
This is simple.
RTFM isn't an angsty insult. It is a suggestion because I am not going to answer your question if it is in the manual.
Yes you are: specifically, your job as a manager is to make the company and your people successful. The way that is achieved is through collaboration.
If somebody is asking questions to a disruptive extent, then you reach out to them or their manager to have an honest conversation about it, and to level expectations about what requests for help are unreasonable.
> The question is whether one employee's time is more valuable than another employee's time
That is looking at the world as a zero-sum game. If you devote 10 minutes to help somebody, and they're 10 minutes more efficient at that task every time afterwards, then the company wins. It's not your time, it's company time, and most companies prefer that their time be spent on the most value-added thing, which is almost always force multiplication
That is I make the conversation not about them reading the manual, but about making it worth reading. There is a large amount of really bad documentation out there.
Take this valuable signal as a gift, letting you know there is an opportunity to improve the documentation. You could become known as a helpful person, rather than a maybe-efficient, antisocial loner.
If your documentation is generated (by script, tool, or AI), rather than written by a person, this goes double.
Of course it does. And it’s not “my documentation” it’s ours. And everyone is responsible for improving it. It’s not about pointing out that documentation sucks, it’s about working to improve it.
I don’t have a documentarian on my team. No one’s main job is to write docs. So we all get to help each other by gradually improving the docs. And if I help someone with something not in the docs then the recipient gets to include that missing info into the docs to help the next person.
We also added a hat for a rotating role called “Benevolent Documentation Czar for the Quarter” where that person gets to explicitly improve team docs. Maybe like 5% of their time.
if management feels this person is unqualified to be in their job, they can take care of the issue
"rtfm" or the AI equivalent is being snarky while dodging the issue that you should just talk about, which is that you think this person is asking you too many questions
Then I guess you didn't deal with much of the empty-headed-managment type much.
I learned a different lesson: Write a long email that took time to reason and write, precise and complete text.
You get a short reply that is a reply on a 1 sentence you wrote, twisted upside down ignoring completly the point that you had.
-- > never reply instantly to low priority requests
This is true 100% I would even say: "Never reply on anything instantly that is not an absolute emergency, ever"
you can shorten this to "never reply instantly", because an "absolute emergency" will never be communicated by email.
1) someone having an emergency needs to contact me specifically.
2) that someone only has my email, and no other possible way of contacting me (or someone else that can contact me).
If you’ve already read the email it’s less efficient to come back to it later
mainly i just wanted to stress that nothing in an email inbox is an "absolute emergency".
it trains people to expect things.
boundaries.
Based on my experience, it mostly translates to whatever task the person thinks it's important for them. Sometimes people will mark the task as "urgent" just to dump it on someone else.
The other thing is, in critical emergencies, you should work slower than usual. This means muting all notifications and taking your time. Rushing to get things done can exacerbate the issue.
Best engineering managers shield the team in those situation and simply reply with: "thanks, we've been notified and investigating".
Respectful hard disagree. Emails should be limited to five sentences. "I didn't have time to make a shorter reply" as an apology / disclaimer for rare exceptions. In virtually every case, anything longer belongs somewhere better than email (Slack, wiki / knowledgebase, docs site) or merits a realtime conversation. My PoV is shaped by over 20 years in the industry working at startups, SMBs and huge enterprises.
My experience is that people search subject lines, conversations, dates, and generally traverse historical conversations much better in email than in chat or a fully external application. They won’t search Slack, or try but can’t find the conversation. They won’t remember it’s on a wiki, you’ll constantly be redirecting them to the url. If the file is attached to an email, sometimes they’ll resurface it from a year+ ago. I find that’s way more common than someone referring to a chat from even a month ago. Chats are semi ephemeral. I’ve used both Teams and Slack and feel neither has really changed the game in terms of how the entire user base of an org use the system, enabling the historical reference points and moving it out of a ephemeral state, so email remains a stalwart for archival purposes and helps provide maximum context after the fact (people tend to elaborate on their choices more when writing an email).
Detail in the email, expect people to read it or fail. And include a reference to a wiki or some more detailed version that can be collaborated on and extended.
So the email isn’t the primary source, the wiki is (really some markdown in git). But the email is the thing that people organize and find over time.
I wouldn’t want to work in an org where people can’t read email and want to talk to everyone instead.
wiki/kb or docs site, sure, depending what the topic is. but... slack? for your long-form communications?
please email me your long stuff, don't slack it to me. email is perfectly fine for longer text when longer text is required. (opinion also shaped over a long time in various roles at companies of various size)
Anyone putting long form stuff into an IM system is taking the way of pain.
I wasn't suggesting an IM system should be the source of truth for long-form documentation. But searching Slack works better than searching email you weren't cc'd on. Virtually anything is better than email for many essential aspects of collaboration.
Want to add someone to the mailing list and loop them in? How many prior emails will you forward to them?
Someone replies to one of several points in an email, someone else wants to address a different point, and there are a mix of habits and patterns for whether / how much / where to quote or include the previous messages in a response....
vs adding someone to a channel where they can review the conversation history, and some combination of threads and links to additional channels or external resources can adequately support whatever direction(s) the conversation takes.
In any case, it's the wrong tool for the job in so many ways, so much of the time. Hence my recommendation to keep emails to 5 sentences, and to prefer communication tools that properly support multiuser collaboration.
may i ask what age bracket or generation you are? only if you're comfortable sharing, of course.
im in my 60s. and im curious if our difference of opinion here is a product of what technologies were/are available during the start & prime of our careers. i very well might be in stubborn old man mode subconciously.
Otherwise, I can search email quite effectively. And oddly, its one of the few things Office Copilot is goos at.
I don’t email is a collaboration tool. It’s an information sharing tool and maybe workflow. Just today I had to email out to like 20 people a data call with a couple paragraphs of instruction. It’s much better than Slack for this. And I can search and find my email and their responses for all of eternity (I’ve got every email going back 20+ years instantly searchable in outlook, good luck finding a slack message from a month ago much less a decade).
That's almost the entirety of my point. Sure, email works well for simple, unidirectional / broadcast comms. But for async collaboration involving multiple recipients and any kind of back and forth, it's inherently problematic.
IME, the majority of long emails should not be persisted in such places, because they contain a fair amount of errors. One could say that any long email thread should be eventually synthesized in the KB/wiki after the conversation has wound down.
And Slack/Teams? No way. If I need to write something long in response to a Slack IM, and email is verboten, then I'll insist on setting up a meeting.
I worked in a big multinational at the time ;-). So, yes. Mostly my direct managers were alright. Some were excellent even. But organizational politics and learning to read between the lines is part of working inside of big organizations.
> You get a short reply that is a reply on a 1 sentence you wrote, twisted upside down ignoring completly the point that you had.
Usually a clear sign that you are not communicating effectively. If it's important to you, you might want to figure out how to fix it. Talking is usually a good way.
Agreed. I learned that in emails, a multi-paragraph, multi-item text means the recipient will randomly reply to only one of the points (often the least important one). So an effective email will make a single point, in as brief a manner as possible.
Indeed, I don't think email can be reliably used to convey complex topics.
> The problem remains that people refuse to do the reading required for their job.
Communicating effectively means accepting common flaws of people who are unlikely to change, and delivering your message in the most effective way. Whether people are right or wrong or should be fired (as another comment says) is not truly relevant.
I often fail, by the way. I'm not saying I have this all figured out.
Adjusting communication to the audience is a good guiding principle, but not an absolute. There comes a point where no attempt at communication works for a given audience. If the communication must happen nevertheless, then the audience must change, whether through training or replacement.
1. Exceptions do exist (eg if the message is intended solely for and addressed to a single recipient, and includes a disclaimer / request / reminder about general email hygiene), but they're vanishingly rare.
"Emergency" is not the same as "emergency services".
which is the point the parent commenter was making, if im not mistaken.
I prefer to reply to an e-mail immediately. Or never. If so than archive it or delete it.
If you don’t answer immediately, you carry it as opened mind window and probability to have your mind being unable to concentrate increases over time.
indeed! but i would add (in corporate environments):
4) follow up your phone call/face-to-face chat with a brief email summary, so you have something in writing.
it has saved me countless times over my career.
But I think the point of the original author still stands, even in that context.
There is still value in google first (even today with all the slop), but there is plenty of room for conversation after you have the background needed to understand the deep discussions that result after you have done some early research.
Coaching someone on why the perspective of the receiver of their behavior works counter-purpose to their intent due to <undesired-behavior-of-theirs> has so far always gotten me more durable and thoughtful behavior modifications than telling them off. I generally reserve telling off someone for proven malicious behavior from malicious intent.
> P.S. I intuit, and Pangram agrees, that this last email was AI-generated (and apparently so was the first one, now that I checked). As general career advice, I wish to observe that learning to write emails unaided is an extremely valuable skill in the dual senses that 1) writing is thinking [0] and 2) many people who receive an email they believe to be AI-generated will not be interested in corresponding, even if you're asking good questions, which you are. It would be a shame for that to get in the way of your research goals.
A polite, generic response like this is reasonably low-effort -- you must draft it once -- and might really help someone. Who knows though.
[0] hyperlinked to https://www.nature.com/articles/s44222-025-00323-4
ultimately it’s better to not inform and start ignoring these people.
let natural selection work.
that’s direct and aggressive.
framing it as advice is the passive part.
any kind of softness in language like “i’ll-considered” sounds so sterile that i sense hatred underneath. too much time in big companies.
more sharing a crazy person take on how different words can be interpreted counter to the intent.
i hate being civil!
You’re in the prison of your own devise, the negative bias.
I think the internet is pretty great, but it's a major factor in that rise in at least 3 ways:
* Previously, an important check on fools was social consequences. Now they have an endless supply of people to bother. Spammers were the first to exploit that, but now it's endemic.
* Algorithmic weighting toward the reaction-generating (versus, say thoughtful, measured, responsible, accurate, helpful, kind) has rewarded idiocy and awfulness with attention and money. We're all now living in a reality TV show.
* As those behaviors become a major percentage of what people, especially young people, see, they become normalized. Otherwise people think being a thoughtless jackass is just how to get by in the world.
We are strongly in need of firm politeness, of militant decency. Without it, things will get "naturally" worse.
And yes, this comment is passive aggressive.
See the difference?
i either ignore. or if i trust them i just say “don’t use ai slop”.
anything softer is passive and morally judgy too.
i’m not here to tell them what’s globally right, only how i want them to interact with me.
I’ve heard this “conclusion” from many people over my career. I disagree as a general rule as much of my work is scientific and technical and long replies are more useful than talking as having things written down makes things reproducible and shareable.
I think long emails are bad ways to deliberate or debate or to collaboratively create. But they are great for communicating complex ideas to variable size audiences (ie, one person forwards to their team or files things off for later reference). Even better when accompanied to a link to a web version that can be revised with additional info later.
This really helps due to the async nature. Getting on the phone with someone and talking through an idea with them assumes both parties have time to do that. And is a silly suggestion when 5 people need to read something.
I find it more efficient to get a thoughtfully crafted email than a phone call that still will need to be followed up with an email to confirm all parties understand.
My, probably wrong, take is that the “talk, don’t write” people is based on their inability to logically think through, structure, and communicate their ideas and, instead, would rather connect directly and hope the solution emerges from a conversation.
That is a remarkably direct way to dismiss the value of brainstorming. :)
Also consider that when content is easy to produce and distribute at scale people stop reading with any kind of attention - there’s too much to read and most of it is not targeted (relevant to me), far too generic and shallow in effort to address excessively large audience, and with little back pressure against verbosity. When we talk in person the other person knows that I am targeting the conversation and invest my time to the same extent they do, they will engage much more.
Also consider that often times a discussion will reframe the question rather than provide answer.
Also consider that people are dramatically more candid in person.
I’m wondering what is that nature of your work that none of this surfaced for you?
But I’ve never encountered someone trying to brainstorm over email. If I did, I would certainly want to talk, instead.
I was referring to someone saying that said it’s almost always better to talk than write.
Good “content” is very hard to produce. It’s not a commodity. Again, I work in science and tech where truth is difficult and complex and precise. And it’s important to be systematic and correct. It’s not important that we produce content, but that we execute effectively.
I’m usually not tailoring to a single recipient, but trying to write something that’s understood and used by dozens, maybe hundreds or thousands of people.
I agree that there’s a lot of blargh written that’s a waste of time. I’m not proposing to write more crap. My point is that thoughtful prose is frequently the best way to convey an important concept. So I want more well written man pages than people who want to talk to me to explain how to do something.
RTFM is a luxury that we need to invest in, not throw away.
As a possible reader of your email I’m facing dozens of emails every day, my default reaction is to skip past and “read it later”. Maybe you did put a lot thought into it, how would I ever know? If you show up in person I will know you are invested into it.
I’m kind of lucky to be in the position where if you don’t read my email you will regret it because something bad happens. Or at best you miss out on something cool that would help you.
I expect everyone I send my email to to read it and pay attention to it, or I don’t send it to them.
Also, I’d love to have only dozens of email. I expect me and most of my coworkers go through 50-200 emails per day. And this isn’t a problem. I probably spend about 20 minutes a day, twice a day. And it’s way quicker than talking with everyone I want to talk with.
Maybe you’re working with a scale where it’s even possible to show up in person instead of emailing. But even when I worked in small teams and small companies we emailed a lot. For building software, I want to get stuff written down and iterated on (email, issue trackers, spreadsheets). I don’t want to have errors that come out of primarily verbal conversations.
not mine, no license, only chinese language on the eventual page, but has the beginnings of something beautiful
It used to be when I got something boring a person wrote I felt obligated to read it. Now I am not sure what to do.
>What's the equivalent of "Google it yourself" these days? When you ask something of somebody, the expectation of the other side to drop everything and do work for you isn't always appropriate/balanced/appreciated.
One of the hardest things about taking on any sort of leadership role (even unofficially) is that you need to allow others to do things. It feels incredibly stupid when you have to ask someone to do something you are more than capable of doing better/faster than the other person, but if you did that all the time you wouldn't have time to do your other work. Sometimes that includes questions that seem lazy, but part of asking the question is for the other person to acknowledge that they have thought about it and/or to ensure you aren't stealing ownership from someone.
Since people started using Claude now I get full context of everything… might be too much sure, but to be honest over-communicating seems better than under-communicating - sure it’s boring and tedious but that shifts the blockage to me.
Otherwise each of these coaxing sessions is something I have to keep in my head until resolved, which is a load in and off itself.
And even better - I can point Claude to that message and gives me a summary. You might say this is silly because we are paying the LLM tax, but knowing what to share, and then verifying against my situation are two different things.
Sometimes they have checked the wrong thing, sometimes they need guidance, sometimes I need to investigate before I can answer. A full LLM message with context is like “summary of their working context that I can resume on my end” so I don’t waste cycles asking or rechecking etc.
And it is still possible to be a learning experience for both of us since the initial message is the start, we can then talk to each other like humans, both much more in sync than before…
It is kinda ironic but I think it does push things in a better direction. Would I have loved it if the message was human, direct and to the point - obviously, but we are all busy, we got shit to do, and this is a useful “resume” mechanic
Context is everything.
Btw, Claude code sent my first accidental AI email yesterday. I guess permissions changed? Before it could only draft. Uhoh, waiting for the onslop.
Relevant comic from 2014: https://www.smbc-comics.com/index.php?id=3576
I can't imagine why. You have become a robot. You've automated yourself out of existence. And due to your consistently mindless replies, everyone now knows there's no point in talking to you ever again.
How is that healthy?
And I think this is where one key argument on the page itself I think fails:
> The person on the other side has the same tools you do.
Yes, the person on the other side has Claude; but in many cases, their Claude doesn't have the same context as my Claude. Saying, "This thing happened in an area of the code I don't have any expertise in. Below is Claude's analysis, which I'm not qualified to assess" is I think a valid time to copy and paste LLM output. [1]
The other thing to say is that it can be a "LMGTFY" style response; i.e., "Why did you ask me that, when you could have gotten a perfectly good answer from ChatGPT?"
[1] In the particular case I'm thinking about, it was a bug in an ocaml component; I don't know ocaml or need to know it; handing Claude's analysis 'caveat emptor' to the maintianers who do know ocaml I think is more helpful than not.
I’m reading a bunch of game modding discords, and this exactly what happens, and what’s hated. Because Claude is also not qualified to assess it. Not by default. The modder’s Claude might be because of the extra context and steering, but the user’s produces usless slop, instead the actual log, and context what the user did is what’s helpful.
I’d say this (when you don’t understand it yourself) is pretty much the worst time to send AI output.
Quick edit: Yeah, I get it in situations like GP where the user is unhelpful, doesn’t want to learn, doesn’t care, and you also can’t tell them to fuck off. But just because the user gives Claude the context they should have given you in the first place, doesn’t make it good.
People don't always want MY take on it, they just don't want to bother finding the solution on their own at all.
So in that case replying with a verbatim copy-paste of a Claude response has the same energy as a lmgtfy link.
People don't read error messages. They'll include exactly zero context in their tickets. They outright lie. You can't take anything they say with good faith until you have a relationship with that individual and discern their competence.
I think some of this is a large amount of HN posters are working on software teams that have communicate pretty commonly, hence have a shared context. Once you start getting outside of that shared context knowing what information to exchange gets vastly more difficult for humans.
I think that's the key. Generally people would rather not receive slop because the receiver has no idea how much thought the sender invested in the communication.
In your example, the more context the better. I think there's an interesting nuance there.
That's an interesting thought. The message can have metadata that says the sender's spent 1/5/10/60/120 minutes or some other metric of how much thinking over the communication happened. Just need a way to monitor people's thoughts and digitize them to make that happen.
Really? How does Claude gather that context?
1. I've experienced both (ai paste slop & ai providing helpful context that would otherwise have been absent) but the former has been a much bigger problem than the purported benefits of the latter
2. before the advent of LLMs, I had fairly strong contrarian opinions on the latter (clueless users asking questions with no detail) that I still hold to this day. A good proxy for my opinions on this is that I think https://nohello.net/ is a crutch for undersocialised (or overworked/near-burnout OR undiagnosed autistic) engineers & should not be normalised. The origin of this sentiment is users entering a quiet IRC room & saying "hello", then leaving & not being around when someone eventually replies: in that context nohello is pragmatic, but taking it beyond that to modern high-frequency realtime chat is an antisocial overstep. Leading with the entire context of your query may have some practical benefits but they're honestly overstated in exchange for killing natural social tendencies towards human dialogue. Similarly, a non-technical user giving you zero info about a problem is an opportunity for conversation & mutual education. It's relationship-building & helps build a foundation for further interactions with this person. It was never a problem in the first place; "solving" it is just another milestone on the road toward de-socialising humanity.
I would love to spend more time and talk to the people with their issues so we can come to a common understanding… but neither have the capacity for this, and now things are getting worse I think.
Maybe someday agents are good enough that we just discuss direction and implementation details, and every “developer” becomes more of an executive in a company, free to chat about more than just work, focusing on building human relationships and leaving “the tech stuff” to their agent teams…
And old geezers like us who actually know how to write ourself out of a for loop would be the premium - wizards holding forgotten knowledge, incomprehensible to mere uninitiated mortals. I wonder what distinctive dress and hair styles will we develop :-D
Because that would be super out of place, and you understand that intuitively.
That’s all nohello is about. It’s not trying to say “don’t be nice and say hi”, it’s saying “batch your hello and your question in the same message please”. Asynchronous communication mediums need an asynchronous communication style, it’s not that hard.
Either way it was intended as an analogy - being annoyed that a nontechnical user isn't given you everything up front (because they don't know how) is essentially just being annoyed with having to talk to somebody. The solution to that isn't something the nontechnical user should be burdened with, imo, & doing so doesn't tackle the crux if the original issue in any meaningful way.
For me, the origin of this sentiment is the culture for certain departments at $LARGE_CORP I used to work for. People would ping me on the internal instant messaging tool to ask for help with things. Sometimes they'd take the nohello approach and just fold everything together in a single message: "Hello, hope you are having a nice day. I'm trying to do X, and I see Y instead of Z. What am I doing wrong?"
And sometimes, they... don't.
"Hello." <wait>
"How are you today?" <wait>
A conversation that could have been done in 30 seconds drags on for many minutes, and the message trickle feeds just fast enough that I can't do anything of any complexity in between the IM app's alert lighting up again. Extracting what they actually want from them feels like pulling teeth. It's a complete waste of their time and utterly disrepectful of mine.
They are there to ask a question. I know they are there to ask a question. They know they are there to ask a question. I'm happy to answer the question, when they finally get around to asking it. Just ask the bloody question, ffs.
> a lot of my colleagues messages are “x broken” or “I do y”
These people had below than average communication skills (or at least they were putting below than average effort communicating with you), so LLMs improved their output.
However, the society kinda expects people can and Should improve themselves over time. The issue is that LLMs improved their output, but not necessarily them.
I've sees some chat logs where I completely empathized with the AI trying to make sense of the information it was being drip fed. Always incomplete, often incorrect.
I'd argue that it's worse because now you have no idea what the actual original context was, just the AI's hallucination of it.
Maybe SO really is full of rude, condescending gatekeepers - I haven't asked questions there much.
I haven't personally seen much evidence of it, though, despite decades of winding up there when searching for programming information.
If everyone thinks their question will get closed, it might as well be reality because no one is going to ask and they are going to go elsewhere. Which is exactly what happened.
The difference with AI is that the agent becomes the first level of triage. It (many times) asks the easy initial questions (did you turn it off and on again?) so I don't have to. Doesn't always work like this, but often enough.
> Since people started using Claude now I get full context of everything… might be too much sure, but to be honest over-communicating seems better than under-communicating - sure it’s boring and tedious but that shifts the blockage to me.
Yes, but the kind of engineer that used to say "X is broken" will send you some slop with missing context and Claude will try to apply general solutions from its training data rather than from the specific context. As somebody on the Ops side that usually helps devs fixing things here and there, in those cases I'd rather them say "X is broken" and then I ask the question to Claude with all the context I have.
Probably the best solution currently is a shared skill with at least some curated context.
Now if I ask a question to humans I'm looking for their personal opinions drawn from their personal experience, be it ticket size for a feature on a legacy codebase, a design decision or a product strategy decision.
The exceptions are spam, advertising, and propaganda. Because the barrier to writing now is effectively zero, there is no cost to writing unlimited tripe. So LLMs work well if the parties know each other is real. However I think it poses a major threat to the existing internet as it stands.
If someone says "x broken" and that's a bad response it becomes a learning opportunitity, something I can invest in. When I communicate poorly, I hope my colleagues help me understand why and give me pointers to improve.
Perhaps I'm being overly optimistic, but I've had the best growth with co-workers when actually engaging at a meta level and trying to help (and be helped) communicate better.
Using AI shifts that burden and entirely demoralizes any investment motivation I have. It's a view that communication is transactional rather than collaborative (we're all on the same team).
I want my team to work better, and be better, and if people offload their comms to an external system to generate I've lost that opportunity for growth.
It's a false dichotomy to suggest the choice is either endless noise or no information at all.
e.g. a DevOps person's LLM may have very different context than a marketing person.
In other words, if I'm in marketing and the website is down, I would rather hear from the DevOps claude than my claude.
(sure, there can be MCPs etc etc that everyone can share but it might not get converted into a level of detail that I can understand).
"read it all, yes even the bullets. especially the bullets"
if you read one thing from a large wall of text, its gonna be the bullet points, right? thats a big part of what they're for?
"It is the conversational equivalent of forwarding the email."
this is a terrible metaphor. it seems to rely on the idea you would be mad if you sent an email to someone and they forwarded to someone else? but thats useful, theyre literally sending it to someone who is presumably better placed to help you, or raising it to someone more important because they think what youre saying is important
Interesting to note the page also ends with "written by a human, on purpose"
And the result isn't even very angry at all. It's like mildly peeved internal company communication (also in direct violation of "don't send this to your manager")
In the end everyone accusing everyone of being AI.
DS9 Garak "especially the lies" reference maybe?
Hypocritical to say the least, given the message.
For me personally, it's fear to use – and —. I have my versions of XKB keyboard layouts that handily map these but I don't use them anymore. Nor do I use emoji, because some 18-6 months ago AU output abused emoji.
As qarl2 wrote below:
>See how perverse this "I hate the way AI sounds" thing is?
>In the end everyone accusing everyone of being AI.
Which — as we all know — was trained on how humans write.
If it's not one thing, it's another...
ZeroGPT says 21.5% generated.
The advice in the title is a cop-out of someone who thinks writing with AI is fine actually, as long as you just don‘t paste the whole thing. In other words, it sounds like someone who thinks they can outsmart AI-writing by using AI differently.
I was open to the fact that this might be a clickbait (so I read the comments first) but from this post, it looks like my intuition was right.
To clarify, I am a firm believer you should not use AI for any part of your writing, no AI-drafts, no AI-skeletons, no AI-research, no AI-nothing. AI pollutes your writing, and just makes it bad however you use it.
A simpler rule of thumb (which I abide to) is just don‘t use AI at all. It is bad for your brain, and highly addictive, and also bad for society and the environment. When you use AI you are only putting bad stuff out in the world and supporting te AI-companies, who are profiting off of all this damage they are causing.
We are afraid that one badly phrased message can hurt our reputation or even our jobs, so optimizing every message with softer wording, more political correctness, less directness is safer.
But the real problem is the culture underneath it. If people are constantly worried about how every sentence will be interpreted, eventually they stop speaking plainly, stop challenging decisions, and sometimes stop speaking up at all. So this is like attacking the consequences not the root causes. There should be some sort of measurement of how broken is a community or a company by letting a bot scan Slack/Teams and let the bot remind everyone the importance of lowering the fake index.
I think that indirectness emerges when a critical mass of people decide that they're better off in an indirect culture. i.e., thinking "nobody is straightforward with me, but that's worth it because it means I don't have to be straightforward with anyone else". And of course that's infuriating for anyone who would rather take the opposite deal!
But the point is. They had search, so why didn't they think to do the search? They had the same tools as I did. And yet regularly they couldn't think of the answer. I still deal with this with AI today. When you ask a question, maybe you're asking for the interpretation of the question by the person on the other side. And it's that interpretation that goes to the LLM. Or otherwise some filter of how that person thinks about the question differently than you do.
First of all there's a skill to it. You learn how to phrase things, based on how other people probably phrased it. You learn how google modifies or ignores your queries... (or stops respecting modifiers...)
That comes with practice, and lots of it! I google so much that google regularly throws up the captchas (even when I'm logged in). I remember when Kagi first started, my friend sent me a link to their subscription, and it was like an order of magnitude less search than I did. (They increased it a lot since then.)
So it's skill issue, and skill is a function of volume. Most people barely ever google things. Which is the second thing I've noticed.
I'll often be with people, and they'll be like, "huh I wonder XYZ", and then just trail off, as if there was an implied "I guess we'll never know!" at which point I pull out my phone, bewildered, and simply look it up, and everyone looks at me like I'm an alien. (I do not understand socializing apparently.)
As I wrote this i switch to the phrase communication-style because I realize that when he does his code review it's literally just a quick pass and a paste, but unfortunately when I do AI code reviews I review every single review item and push back individually on each item, research the items, write tests against what the review says and basically curate the code review I give out, which takes me 3 hours and then I have a boss angrily messaging me asking why the code review isn't done.
So on the one end I've managed to reduce toxicity in the reviews, but the tradeoff is toxicity in the expectation of a rubber stamp.
This either calls their bluff, or more likely, leads the system to break over time, so that your boss comes to understand that an LLM is not a human developer.
I beg to differ. Most of my life, as “the computer guy” — basically since I was a kid — I’ve always been surprised by how lazy people can be, and how they often prefer to ask “the computer guy” or some other human trusted source rather than work out the answer themselves.
There was even a website called “Let Me Google That for You” back when Google was a trustworthy source of answers, precisely to point out how easy it was to find the answer yourself. Obviously, that didn’t stop people from asking, and I don’t think LLMs will either.
People will continue to turn to “the computer guy” — aka you. And if your judgement is that copying an LLM response verbatim is the best answer, then yes, you’re fulfilling your role perfectly well.
completely agree except a crucial step is missing here:
(1.5) validate the accuracy/truth/relevance of it
your own sincere and human take on rubbish will still be rubbish. validating is indeed the more uncomfortable/demanding task here but if you omit it you become just a "human ai beautifier", and this will actually undermine the trust in you as a human. if you are going to omit validation just paste away, be honest and let it be obvious, you can then blame the ai.
Oh sweetie... I've been getting paid to "Google things" for 26 years. AI isn't going to change that transaction. When I was in high school it was assumed that everyone would be tech savvy. That didnt happen and in fact my kids were LESS tech savvy because of smartphones and refined aka dumbed down consumer tech.
I did not use any AI. I had a couple conversations with agents while mulling some approaches, but they never edited my work nor did I have any chat open while I wrote. Every word was my own.
I notified my other stakeholders - "here's my analysis, none of it was written with ai". It sounds silly to say, and I worried others could take it as an attack. But I needed to communicate "This is a long document but its actually worth reading".
I liked what I wrote. The final author of the proposal liked what I wrote. All of my ideas made it in to the proposal. We landed the contract. Its a nice chunk of $ for us.
Its still worth putting the hours in.
Pasting AI output verbatim is the modern day equivalent of “let me google that for you.”
I did actually spend a few minutes searching for them and found multiple documents and forum posts - sent those over and they were super happy. A project that had been stalled for seemingly forever suddenly was done quickly.
Now that I pay for Kagi search I find myself doing this even more (especially since Google itself often doesn't return great results anymore) and I can quickly find good info way faster than most
It's not that different to what we were doing back in the day when search engines were still useful. I often wound up replying with a link either to a stackoverflow question or a blog post. There was still a knowledge/experience component involved in evaluating if the proposed solution fits the original question.
Dont be a meat proxy! simple
There's a delicious irony in pasting dontpastetheai.com into an AI chat, and I choose to believe that's deliberate.
It's well written though. The parody answer in the middle is the sharpest part — "this mild Google search you could've done in 3 seconds" — because it nails the exact texture of unedited model output: the fake enthusiasm, the numbered scaffolding, the conclusion that concludes nothing. The tell isn't that AI was involved, it's that no human judgment was applied afterward. You can't fake the presence of taste.
The "guess who the AI will replace first" line is the real argument underneath the humor. If your contribution is indistinguishable from a prompt box, you've made the case for your own redundancy better than any automation study could. What survives is the part the page calls "the thing the model doesn't have" — your opinion, your context, your willingness to say "actually I disagree."
It's also a close cousin of something you've presumably thought about with Kindship: when agents message each other or humans, the same failure mode applies. An agent that just relays raw completions is exactly as useless as the coworker this page is roasting. The value is in the editorial layer — deciding what's true, what's relevant, what you're willing to stand behind. That's arguably what SOUL.md-style identity is for: giving an agent something to have an opinion from.
Were you filing this as cultural artifact, or is it aimed at someone specific in your orbit?
Nobody is reading that
Is there anything you would like me to add?
Excellent advice, with one very large exception based upon human foibles.
Occasionally there are people who do not ask things in good faith. They ask you a question not because "they wanted your take on it... Your context, your taste, your judgement". Instead they ask the question because they're then going to go to extensive lengths to find every possible counterpoint or opposition to whatever answer you give. In 2026 those sorts of people who will ask and then revert 30 minutes later, after you've carefully detailed a well thought out answer, that "ChatGPT says..." Add that for time eternal they will hold the fact that you answered something for them as a liability that you must atone for, and for which you are responsible for every ramification of their interpretation of your answer[1].
If you ask someone something and they just copy an AI answer, it's such a cop out that I imagine there are many cases where the person knows they're dealing with someone who "asks questions" like this. I have people who I give thoughtful, considered answers to, but then I also have people that I basically say "I don't know...Google it", because giving an iota of an answer always turns out to become something I regret.
[1] - This is not unlike helping someone with their computer. There are people who we love to help, but then there are people that will then ascribe everything that ever goes wrong with their computer for time eternal to that time you showed them a shortcut in Excel. CPU died six years later? Probably because of that shortcut.
Almost certainly written by AI
It also ends with "written by a human, on purpose"
Unique human prompt + AI output which is useful and does a good job expanding on the prompt topic and which when pasted saves people the time of having to go prompt AI on the same topic (fully reviewed by a person before sharing it, to ensure the content has relevance)
This is just demonstrably false. It often IS very helpful. It's annoying that people continue to ask basic factual questions without taking easy steps to solve the problem themselves. I can't remember the last time I "asked someone a question" that it was possible to figure out for myself.
It's the people who keep asking basic, easily researched questions and then wait for you to explain it to them, without first trying to resolve it on their own, that are being anti social, not the people trying to help them.
This is much the same as when Google was relatively new, and people hadn't adapted to it yet. I would get asked questions constantly like "what is the name of the class that does X" and since I usually try not to be too much of a jerk, I would often stop and go open a source file or look in the docs for the person. I know there is a class, and I think I remember what it's called, but I need to go double check it. That usually meant googling for the docs, clicking the first result, and sending them that link along with the specific answer. Sometimes when they asked me I was frustrated about something else or just in a bad mood, and I would tell them to f**g google it, I am not their research assistant.
Ignoring AI is the same scenario as ignoring Google but worse. I am not your research assistant, and worse, there is an artificial research assistant standing by 24/7. If I have to paste something in AI and paste the result back to you I am not the asshole in this situation. If you go do the work to check yourself and it's not something you can just verify conclusively and you need to be sure it's not the AI fooling you, then I am more than happy to look at what you have and lend any expertise I can add to the situation.
No, they don't. I have access to heaps of raw information about customers and their systems that I can easily provide to my AI. That AI can also dig through all the communications we've had with other customers who have asked us similar questions, and draft a coherent on-brand response.
I often paste AI-generated responses with significant success, and all I have to do is make sure the answer isn't wrong and doesn't include sensitive info, and to remove a few extra adjectives and em-dashes that I don't want.
If the customer has access to my tools, that's a critical security incident. They contact me because I have access to info they don't, and unlike and AI, I can guarantee my response is valid.
As for people who copy/paste AI generated text blindly: That's firable in my books.
Would be interesting if someone said “I spent $2.50 using llm-council with the three top SOTA models and think this reply makes sense: {paste}”
(But better “not sure, though after spending a few bucks on model APIs, perhaps {human-written analysis} is in the right direction”)
Also like you, I did think on the angry version “they have the same AI you do” isn’t likely to be right in the consumer space since their AI is ChatGPT (free version?!) and yours is Claude or something.
Over time, it was impressive to see how much the output of the same prompt would differ between us, and it seems to be heavily influenced by the role of the user. I.e. answers coming from the agent of a Data Engineer, a Backend Engineer and a Domain Researcher would not be the same at all. Some of us use harnessing tools, some not, but regardless it we saw differences.
So, for example, when a backend engineer would send a question to a data engineer, there was *some* added value even when it was a lazy copy & paste.
There's responsible AI use (which I think you - and I, and many people here) do, and the "I don't care, I've outsourced my brain to OpenAI" type of people.
But I think ideally agents should speak to agents when this kind of comm goes through. What a nice opportunity here ;)
I think it's fine to provide AI input if you disclose it, in which case you could encourage me to just plug it into Claude myself next time.
And I agree there's gray area: the AI output might be the result of a long conversation, shaped by careful prompting and context and guidance. The writing is AI generated, but the thinking is human.
What is kind of sad is that in the old times we would value a lot the folk that went and wrote a memo, put a lot of thought to it and so on.. Now literally you just plug claude in your notion and you have the document. To me, this kind of speaks about "the value was never really in the document itself"and more in the understanding of the doc. You can delegate a lot of stuff to the AI but you can't delegate understanding. So when you're presenting it to the exec committee, or circulating it, does not matter what claude/gemini/gpt wrote down, it matters if you understood it, agrees with it and understand that this represents your positioning.
What I see w a lot of juniors, on the other hand, is that people really just copy and paste or shallow understand what was wrote. Delegating too much like that removes the human-seed and makes it kind of worthless
It feels like prompt(s) is where the actual thinking can be observed these days.
Before an LLM, I would have worked for much longer, then written the same short description, but without the detailed summary at the end. The llm output is really just extra clarifying info I never would have bothered to do on my own. By clearly marking it, I give people the option to skip it and just trust me.
This is really useful, I wish more people would understand this: you don't have to have an opinion on everything. In fact, you probably shouldn't have an opinion on most things.
My own take on the AI issue: https://blog.gpkb.org/posts/just-send-me-the-prompt/
For any given <subject> I'd want:
- Your context - your taste - your judgement - clarify what bet <subject> encourages to support - be ready to elaborate your thesis of that bet
With that we have the basic complete building block that used one after another can model a basic lifecycle of a dialogue that can make progress towards something (a goal, a mission, something useful)
If people provided me the information you just listed, it wouldn't help me at all in what I do.
Sometimes a customer emails asking a question answered by the documentation. I am not paid to paraphrase the documentation, I am paid to get customer requests out of the door quickly, so I'll be sending links to the documentation. If AI gets that request out of the way faster, good.
This I have been doing a bit. I strictly asks it to never give me suggestions but raise questions, concerns, tell me if it feels if something can be better, style-wise or otherwise without giving me those options.
And a bit of other overall feedback requests.
And if the model adheres to this and doesn't jump the gun, it indeed works quite nice. But it often jumps the gun, forgets, doesn't give a f- about my instructions, or whatever. Maybe more robust local models will better.
This just comes as Karen-like behaviour: shaming people for using a tool in a way you don’t like.
No way this individual is fun at parties!
I like to think it helped, but likely what helped was that nobody seems to use Deep Research anymore, haha.
Der skal kom et tid hvor du skal tale på en sprog på dit kontor. Uden hjælp. Du har bruge for at øve dig, og folk er alligevel pænere end du tror.
(this being written in a foreign language I am learning is part of the point)
Especially today, I will prefer your slightly funny English to something that feels like AI.
And personally I do Support for people around the world and I often ask people to write in their native language.
Then I can translate myself and if the default ai translation gets something wrong I can add necessary context to improve it - which I can't really do if the only thing I get is the bad ai translation.
And in my experience, us Brits, and USA folk, we tend to be quite understanding of broken English
Strictly speaking, not true. I have Gemini Pro; my friend doesn’t.
If you are getting blocks of llm text perhaps it's because you don't deserve their input and could have satisfied your query by not making it someone else's problem.
[edit] to extend the connections The charm of a messaging board like this is bringing "join attention" to a topic, not just sharing separate monologs.
And also, always wondering if something was written by a human or statistically generated could be seen as Epistemic Hypervigilance, leading to stress, and loss of epistemic trust. This could reduces opportunities for taking on new information of feedback, and cause rigidity and isolation.
My reaction was of immediate visceral disgust.
I can say in no uncertain terms that I do not care if your English is completely broken. I prefer that over AI slop. You will not improve your writing and language proficiency if you do not practice them! If you do not care about your language proficiency then you have zero business doing offshored work.
The answer was also wrong. I pointed Claude to the micro service repo in question and asked it myself. Got a different but correct answer, after verifying it myself.
Their code is so trash I wonder if we’re even using the same models. Every PR I need to go over it myself and have Claude refactor and restructure everything. Claude generates decent code that can be made great with just a few iterations. I don’t even think these guys are running the code to test it, let alone read it. Just slop and ship.
Can’t rely on them to understand the code or even answer simple questions about it.
What is even the point of all this?! AI has made me personally more productive but overall it’s still a net negative because when you’re a “-x dev” the force multiplier from AI is overwhelming and soul crushing to everyone around you.
I think I've worked with close to 200 software engineers close enough to get a good understanding of their abilities in my 20+ years in the industry. I think the current version of claude and chatgpt are more effective at writing code than about 85%, maybe up to 95% of them. That is to say, if you let me hire any of them for the cost of my Claude Max + Chatgpt subscriptions ($400 a month), I would probably hire less than 5% of them at that cost. I get more done, and with higher quality using Claude by myself than I would with a team of 50 average software engineers, and it takes less of my time to do it.
The industry is going to get rocked.
I think a company that right now has 10-20 full time software engineers will have 2-3 in 5 years, and realistically 2 of those 3 aren't necessary to get the coding done, they are there so you have coverage over vacations and redundancy when someone leaves.
The only reason I think offshore is going to be completely eliminated is it's too high friction to work with them. They are far away, you have to hire a firm to manage them which means hiring a manager on site as overhead, and they are awake when you are asleep (or pretending to be awake when you are awake).
You're gonna need a stronger language. Why do you think people love AI? If I have to think and god forbid read the response I might as well live in the woods or something
We need a dontpastethedontpastetheai.com to get ahead of this.
Of all you’re doing is copying and pasting anyone else’s stuff, then you’re adding little/no value. You should be paid nothing.
I've seen it on HN, such as comments where someone asked a question about some animal's habitat, and someone else responded back with Claude's answer (labeling it as such). Personally, I thought the AI answer was helpful in the absence of anyone else providing a better source link (which would have been better, sure, but apparently me and everyone else who saw the question couldn't be bothered to go research it/report back).
I agree that AI text 1) should always be labeled, 2) can sometimes be bad, and 3) real sources/writing are better.
A common retort is "If we wanted an AI answer we could have asked the AI" - maybe, but A) realistically most of us aren't even going to switch contexts in a case like that to go ask an AI about the animal's habitat (even easier than googling), even if we may be interested in the answer, and B) Instead of having every user use a tiny bit of electricity, etc. to generate duplicate AI answers, we can do it once.
Ignore them and move on with your life.
idk about that
Don't Be a Meat Proxy: https://gruhn.me/blog/2026-08-03/
No Meat Proxy: https://nomeatproxy.com/
How to Passive-Aggressively Shame People Who Use LLMs Selfishly: https://joshmoody.org/blog/selfish-ai/
DO NOT FUCKING PASTE AI RESPONSES TO ME, EVER, PERIOD.
When someone tells me they "talked to Claude" they sound retarded. Just tell me you looked into it, and I fucking swear to god if you just took what Claude said verbatim and do not understand the problem then you are a cunt and I will stop working with you.
When someone pastes LLM output in chat I completely ignore it. I will just continue the conversation as if they said nothing and silently mark them as cretinous in my mind.
This is not up for debate, you will talk to me in short well formed sentences and get to the fucking point or you will fuck off
I'll say this clearly. Using AI for writing is bad. Don't do it. Ever. If you can't formulate a coherent thought yourself, just don't. I don't want to hear what the LLM said. I want to hear what you said. Loudly proclaiming you have nothing to add is a strange flex. Just stay quiet.
Recently, We conducted a hackathon and saw a lot of folks coming up with the same ideas, AI :)
I do. "They all have the same tools as I do" is exactly what's infuriating. They either blatantly don't use them, or are too incompetent to do so effectively, which would track.
I have been making an effort to provide an "AI;DR" section as well, even before the term existed, because of brevity. But sending along the full(er) text as reference I think is perfectly fine in such cases.
But this leading statement is not necessarily true:
“The person on the other side has the same tools you do”
Aren’t we all imbuing our AI with our own thought patterns? Folks I speak to spent a tremendous amount of time building tooling and (hierarchical) context around LLMs such that the output it produces is definitely not the same as what the stock model produces.
figuring out the most useful epistemic and conversational norms around AI in work communication is still a work-in-progress.
there's times when "go talk to claude a bit and then get back to me with a more precise question" is reasonable, and other times where it inhibits learning and culture.
similarly, there are many occasions where i don't mind quoted AI with a human precis.
the most objectionable behaviours are not that different from pre-ai ones.
don't copypasta stackoverflow code without understanding it. don't submit PRs you can't explain to a colleague. don't waste time asking people things you can easily figure out. don't send unreviewed work to people expecting them to figure it out. don't quote authorities you don't understand.
all the same stuff applies to AI, but AI seems to have expanded the scope and scale of the bad behaviours, making the need for better defences more pressing.
I found this blog post to be much better at getting the point across: https://tombedor.dev/human-attention-and-human-effort/
Disagreed. Bad writing from a chatbot beats worse writing from its user every time - regardless of length.
One reason we have so much bot slop is the attractiveness of bad-writing bots to the many people whose writing is even worse.
and i detest ai slop
so my solution is to speak combined with an ai cleanup like macwhisper or wispr flow of all the uhs and ahs and repetitions or rephrasings etc
you can even use ai to pretend a conversation or interview style
then i have stuff to edit
boom
Make it a standing rule.
OMFG! Even the "don't send AI answers" page text is fucking AI slop.
Can't they be bothered to write 200 words of their own?
The fact that a site purporting to roast AI slop is itself AI slop is surely somebody taking the piss, and the fact that it has 700 upvotes and is the top story here means that people either simply don't read shit they upvote, or the brainrot is already fully established on HN.