upvote
At my current job I just asked a coworker in chat if the reply was his or LLM. He quickly owned up and the incidence of brokered LLM messages in chat dropped signficantly.

Unfortunately people are being encouraged to use LLM in docs, emails, and presentations so it's probably a losing battle.

reply
That's a better approach than what I did, and I'll use that going forwards.
reply
We had an exec show how he was using AI to write interview feedback from his very sparse notes, in front of a few thousand people.
reply
We have a LLM bot in chat that analyzes bugs when.you drop in a link and anything else if you ask it directly. It's actually quite effective: all the slop is now localized and can be ignored/blocked easily, and people no longer feel the need to copy and paste slop themselves.
reply
I've been thinking about these kind of interactions a lot lately. I don't know what it says about human communication patterns but I'm not super excited about it.

I've been guilty of acting as a middleman for review comments from difficult coworkers. I farm out my interaction to avoid the confrontation, mental exhaustion, and to avoid negative feelings I got from those reviews. I don't mean 'my work is being challenged', I mean people who: delay, nitpick, blow up architecture after not participating in design, assert false info as truth and require you prove them wrong, require 10x evidence of everyone else. Not people who are like 'Hey, I don't understand this code can you add some comments or lets hop in a room for review'

I've also been on the receiving end. I like to think I'm NOT one of these toxic reviewers.

Now, we have Business Owners and Managers who are now even further away from engineers. Important meetings have 0 engineers in them ( a couple people who think they're engineers because they were for 5 years in the 80s ). Output is an AI generated transcript that gets copy/pasted into copilot to generate jira tasks by the Product Owner.

It seems like we (humans) really don't like working with other people. Or, maybe we like working with people who can advance our careers and want to automate interaction with people who cant?

I think this effort by Business Owners to set up an AI wall may be the closest I've seen to 'digital classism' (maybe there is a better term for it?).

reply
>Or, maybe we like working with people who can

https://croissanthology.com/earring

We like whispering earrings. We'll follow a low friction happy path even if it leads us off a cliff.

reply
Would you be OK with someone looking up the answer on the internet?
reply
You mean like when juniors would copy paste Stack Overflow without verifying or even understanding it?

No. We hated that too.

Actual research is, of course, fine.

reply
I see a possible misunderstanding on my part here.

My experience has typically been that engineers respond, "Claude said..." and then say what they think about it. If people are really receiving just a straight dump of AI response from their colleagues, then yeah, I would be disappointed too.

But, I have no problem receiving from a colleague, "Claude says ..., but I don't know what to make of it." The embedded space can be really hard, so I don't fault someone for feeling clueless about something. I feel that way every time I get handed something new to work on which I have zero experience with.

reply
We're all complaining about the low effort verbatim dumps, not disclosing Claude usage as part of an earnest attempt to help.
reply
I did a similar thing and I got reported to HR.
reply
What was the outcome of the report to HR?

I think if you work somewhere that this is the result, you're unlikely to have much progress challenging anything directly. In order to actually have change you need to be able to challenge and stamp out these sorts of behaviours.

reply
> What was the outcome of the report to HR?

"Do not make negative assumptions" basically. But also a drop in how often and quickly I did reviews.

reply
[flagged]
reply
This smells an awful like a post-hoc justification of why you can’t answer it yourself.

If you _could_ do that and hand the information to Claude, you could also just…give the person that information directly, and avoid wasting their time and the tokens.

reply
Uh? I don't understand, are you saying that instead of asking Claude myself I should write a question for other people to do it? Which is by the way only the initial part of the work I described?
reply
You can do whatever to get answers to questions, but don't just forward said answers to other people, tell them in your own words / interpretation. Prove that you understand the information you've been given instead of being a meat proxy.
reply
You know what? If you post a question and I get a good answer from Claude, it means you haven't even been able to do the basic job of asking yourself. Not only you weren't able, you also expect me to reformat nicely the answer so you don't feel stupid reading the LLM text you should have generated yourself?
reply
You can also respond ‘I don’t know’ instead of doing the equivalent of responding with the first page of an ‘I feel lucky’ google search
reply
This is a very good analogy. You have to do way more effort to make sure it's correct than if you have asked it yourself with a controlled context. I generally don't want to feel lucky when I ask questions to coworkers.
reply
> Then do it and don't ask questions.

I'm not sure if you're trying to be funny, but this is such an unintellectual take that it veers into caricature. Props if that was your purpose.

reply
No, it was a suitably nasty reply to someone who boasted of being nasty to colleagues trying to help.

If your question can be answered by Claude, and you didn't ask Claude yourself, and when other people do it for you, you complain that you could have done it yourself, then my honest reply is that you should just shut up and learn to use Claude.

reply
The entire point of the post is that people want human responses regardless of whether they can ask Claude or not (and maybe they did and didn't get a good response so that's why they're asking other people).
reply
Every question can be answered by Claude. Not every question can be answered accurately by Claude. When Claude answers inaccurately, it doesn't tell you that it's doing so, so every answer must be considered "maybe inaccurate".

In business, a lot of the time, accuracy matters. If you want an accurate answer, and there's a human who knows, don't ask Claude.

reply
> Not every question can be answered accurately by Claude

You are assuming that people that send you a Claude answer would send you any Claude answer. This is your assumption- basically a negative prejudice towards those who use LLMs- and has nothing to do with sending a Claude answer per se. Do you understand the difference?

reply
I do. I'm pretty sure that at least some humans will shove any Claude answer at you. And some won't.

You are assuming that I am making an assumption. Your assumption is wrong. I am recognizing that some humans are lazy and will do the minimum, and the minimum is to type in some query and copy-paste the results. If you like, yes, I am prejudiced - prejudiced against that behavior.

(I mean, I still trust a human who knows their stuff more than I trust Claude...)

reply
But then we agree. Some humans will behave that way. Then direct your anger against those who do so, instead of against those (the majority) who ask the questions, read and evaluate the answers, and forward them to you only if they think they might be correct and helpful.
reply
If I've asked *you* a question, it's because I want to know what *you* think.

I don't want to know what Claude thinks.

If you don't know the answer, say that. "I don't know" is a perfectly good answer, albeit one which I might follow up with something like "Well why don't you know? This is literally your job to know", which is kind of what 90% of my emails at work for the past three weeks have been.

reply
Out of interest, do you pay people extremely well to work with you?
reply
"I don't know" is an extraordinarily powerful answer, but maybe it's getting ruined. I'm not sure it can be entirely ruined.

In a human interaction, when used confidently, it means 'I'm so authoritative I don't even have to front. I don't have to do the slightest thing to impress you. My foundation is so secure it doesn't need defending or justification at all". Also, 'that turns out to be outside the massive quantity of stuff I do know, how interesting'. Think Holmes, perhaps erroneously saying 'Now that I know that, I shall do my best to forget it' on the grounds that his work wasn't helped by the fact.

It could get ruined by an AI normalization of the idea that 'everybody should know literally everything, because asking an AI is the same as knowing and then you know the thing, having been told it'. Plenty of money is now marketing this position to the world, and so it's possible 'I don't know' will become ruined, a shameful response.

I feel like that may not happen in practice, though, because that pro-AI position is what used to be called 'pants-on-head R-slur' and I feel like the people who buy into this extreme position will make SUCH a disaster of things that it will tend to make that position, instead, the shameful response… and 'I don't know' will remain a valid signaling mechanism to convey 'I'm so authoritative I don't even have to front', all the more if it's more of a shocking response to get.

reply
I definitely appreciate being told "I don't know", but I've found in business settings that people expect to get reprimanded for saying that more than a couple times a year, and they refuse to say it.

They will invent crazy workarounds to avoid those 3 words.

Even if I ask "Do you know X?" they will not just say "no" most of the time, unless it's really outside their wheelhouse. If they think they should know it, they'll spit out statements worthy of an LLM.

And I admit, I felt the same at first. The first few times I told the CTO (my boss) that, I cringed. But he just told me to put aside time to research it. He knew I couldn't know everything, especially about things that we hadn't done before.

reply
We are talking about questions in the context of reviewing a pull request. If I ask you a question about the code, and you have to ask Claude, or otherwise cannot tell me why the code that you wrote is the way it is, that is strike one of two.
reply
Good catch to ask about the code &emdash; if you can't actually explain the code you wrote, why do you expect me to explain to you?

;-)

reply
Most people who post LLM output don't do what you do though.
reply
If someone asked you a question, it's most likely because they value you. Whether it's your skills, your domain knowledge, or something else, people ask you questions because they know you can provide some value. When you give them a Claude response, not only are you being rude, but you're devaluing yourself in the eyes of those around you. They learn that you can't be relied upon any more, and so they stop asking you questions.
reply
Asking the LLM a question _is_ a skill, just like googling. The problem is not asking a LLM, but pasting the 300-chars response without any context.
reply
"I've been asked to confirm that our random number generator satisfies section 14 of the Foo-Bar Accord of 2021. I'm fairly certain it does, and that foorandom.c is a good choice. Assuming I'm broadly correct, write one short paragraph that would confidently assuage anyone of concern without raising any new concerns. Cite the specification and link to it. Afterwards, for my benefit, write a short bullet list with any non-trivial reasons I might be wrong."
reply
In that case a better answer would be "I'm fairly certain it does satisfy it, but in any case I asked Claude to double check for any issues and it didn't find any unsatisfied requirement". That's the information that matters, and it's probably faster anyways.
reply
There is no skill whatsoever in asking the LLM a question.
reply
If I asked a question to Claude, it's because I couldn't answer it myself, but also, because I used my skills and domain expertise to explain clearly the problem to Claude.

And if I paste back to you an answer from Claude, it means that I have read and evaluated what Claude wrote back, and I have extracted the part of it that most clearly contains the answer as well as the diagnosis- if I gor anything relevant.

Of course I could instead pretend I came up with the answer myself in 5 minutes thanks to my superior analytical skills, and write down two short lines that point to the solution without any explanation. Would you find it better?

reply
But that’s not what happens - people are posting the question into Claude and pasting back saying “here’s what Claude said”. There’s 0 value add in that - I don’t know if you agree, disagree or have even read it. My experience is that most of the time it’s just verbatim,

> write down to short lines that point to the solution without any explanation. Would you find it better?

Yes. That’s why I asked you and not an LLM.

reply
> hey, I don’t know the answer to your question but I went through Claude and got this response. Let me know if that answers your question otherwise we can dig deeper together

That’s all you have to add to make this into a much more pleasant interaction

reply
I agree. However, since I work in a small team and everyone knows me, I don't need to provide this entire context every time. I usually just prepend "according to Claude ..." and try to provide a one line recap of the main point. Then paste the relevant section of Claude's answer (two to three paragraphs).
reply
Your response misses the words most critical to this interaction: "I don't know"
reply
deleted
reply
You took all this time to reformulate the question, add the relevant implicit context, and do back and forth BUT you did not have time to reformulate the answer to something human readable that YOU own and can take responsibility for? This does not make sense to me. Nobody said not to consult an llm while formulating an answer. But if you give an answer, own it. Don't throw slop grenades.
reply
…yes?

The amount of text isn’t equal and the benefit is primarily that I can push the LLM into a basin you don’t know the keywords for — something that takes a few minutes to type a couple sentences and me scan the results to make sure they’re roughly on track. And much, much less time than me trying to (re)write a treatise on a topic.

Dropping you an LLM reply is the equivalent of dropping you a link and telling you to learn about the topic — and appropriate in roughly the same circumstances.

reply
Why would I reformulate what Claude already explained in a way that I find satisfactory? Only to give you the impression that I came up with it myself?

> if you give an answer, own it. Don't throw slop grenades.

If I send you some text copied and pasted from Claude, it means that I have read it and I think it's worth for you reading it. In this sense I'm already owning it. The problem is not the pasted answer, it's your assumption that a Claude-generated text must be "slop" that I haven't read.

reply
I would at least expect to share your own take on it, or what you think is interesting or important in sharing an answer. If I am sent an llm copypasta I have no idea of the amount of intentionality or the amount of effort is being put into this. I have been sent many times "I got this from claude, did not have time to look at it closely, maybe you find it interesting". No thanks, I can and probably already have tried with claude already (did not say exactly that). This is the kind of slop grenade that is basically noise, and when you slop grenading me without telling me what you think yourself I have no idea if it is worth my time reading it or not.

I feel like these things are pretty trivially obvious. When there is noise, you have to make sure your signal is not taken for noise, if we assume that your intention is to produce signal in the first place.

reply
I do not think people should stop asking questions. What is the point of teams / departments / orgs ?
reply
devils advocate; why didnt you ask claude then? stop wasting people's time when you could get your own answers.
reply
Because I wanted to ask a human who’s job it is to know, instead of an AI that might make things up and waste all our time.
reply
Why is it their job to know, but it's not your job to know?
reply
Organisations typically divide up responsibilities between different teams, and also between different members of each team.
reply
So, people are using AI to answer questions about tribal knowledge? Is that really the assertion which is being made here? Because I find that hard to believe.
reply
Different roles and specialties? Different work history?
reply
So, you have different roles and specialties, but you know for a fact that this person should know the answer to your question. I think we're speaking past each other, because this doesn't make sense to me.

I don't presume that someone will have an answer I'm looking for. I'm pleased when they do, but often I just get breadcrumbs, and I have to solve the puzzle for myself.

reply