Don't be a meat proxy

Is this the new "let me google that for you"?

Nah, I disagree let them pay for my query dude. I want all the tokens I can get.

PS: Ideally set thinking to max and research mode or whatever anthropic calls it.

Yeah, I just ignore people who do that. I don’t care what Claude or ChatGPT said. If you cannot be arsed to write yourself then I cannot be arsed to read it.

> NATS control-plane events: stream leader election / R3 quorum re-form during pod churn.

this is perfectly understandable?

I’ve been saying for a long time that most fights in your personal life are a proxy war but I never thought of a single person as being a meat proxy. Nice.

Of course your grumpy coworker might then respond: "So if you can ask Claude, why did you ask me then?"

Claude told me to tell you, you’re right

This is mostly a problem with the product. It should write better. But maybe the priorities now are not on this part.

thank you for this post, this is absolutely happening i just did not have words to put on it and it is festering.

This conversation is important but mostly they highlight transactional and most visible examples.

I can tell you even at one higher level of abstraction, this is happening a lot --

A business analyst supposed to ask discovery questions.

A project manager supposed to list risks and assumptions and a timeline.

A consultant expected to outline a new proposed solution and the pros and cons and trade offs.

An architect supposee to present and defend a solution design -- and the change review board supposed to ask questions to validate it.

All of them are lobbing slop grenades at each other -- just packaged in normal / human-looking formats like excel and word and powerpoint and email -- but slop nevertheless with minimal to zero value add by the human, thanks to proliferation of copilot (among others).

We are hastening the rot.

This is productivity theater.

This is where some of the 10x engineer and AI taking jobs is happening I am afraid.

"That's great for you, I'm not going to read it" is a phrase I have had to use.

I agree. How can I see the rest of your blog?

Thank you!

Finally I have a word I can use for this

I am sure this will lead to can you dumb this up prompt which will miss the important details

We added this communication rule to our company handbook: "If you are asking for human attention, demonstrate human effort" (coming from https://news.ycombinator.com/item?id=48497609) and it's working well so far.

This is the new "Let Me Google This For You".

We're seeing this everywhere now, people generating thousands of lines of documentation or PR feedback in seconds, then dumping the actual labor of verification onto their teammates.

> Reading AI output is extra effort. It's verbose, frequently contains all too plausible nonsense, and is increasingly jargon dense. I recently got this sentence from Claude: "NATS control-plane events: stream leader election / R3 quorum re-form during pod churn." Jesus. I had to lookup almost every word to make sense of this.

I'm always puzzled when I see this kind of commentary. How are people using LLMs that they find this type of thing problematic? Just do:

- "It's verbose" -> prompt "Explain briefly"

- "plausible nonsense" -> Yes, it happens, but less frequently than with older models, and arguably far less than your average human. Mitigate by prompting "Run a search to check that X, Y, Z is correct".

- "I had to lookup almost every word to make sense of this." -> Does this really happen frequently? It doesn't to me. Mitigate by prompting "Breakdown that sentence and explain each term in plain English".

This is an interesting problem. A similar approach that worked for us was to only load the tools the agent actually needs for each task.

You either get the AI notice or you don’t. The AI output will still be posted.

It’s a meat to meat courtesy during this transition phase.

Too many damn people are meat proxy

Claude's not a person. Claude is just a tool.

It's my very same exact opinion and as an EM I'm trying to push this everywhere (team, peers, upwards) but it feels like speaking to a rubber wall. It starts to be exhausting and infuriating. This is not an "LLMs are bad" issue, this is a "humans are lazy" issue. (shit, I wrote the it's not X it's Y construction!)

Friends don't make friends read slop.

If I have something Claude wrote then I will offer the LLM output. But I generally prefer to at least give the option.

I have seen conversations where it has boiled down to an exchange of A's Claude responding to B's Claude with a retort and fixes from A's.

I've gone another route when I found myself simply replying with an investigation made from an AI: Actively encouraged my coworkers to ask the knowledge-bank AI I've setup instead of asking me. It has read-only credentials for all our environments and access to our non sensitive codebases, so it can answer any question much faster and accurately than I can. It works too! We use grafana less, we get more accurate bug reports, we understand the bugs/issues far better.

The "problem" here is that AI is too great at providing answers, so lean into it instead of fighting it. It won't go away.

At least "Claude said" is honest. What's worse is when people send you slop and pass it off as their own work and expect you to take it seriously.

eventually two people will just be forwarding AI outputs to each other, each assuming the other one read it

In these situations I always use AI to respond

The question is what is reasonable. This is more how I work for code and I just tried for a document.

i) Prompt (Human input)

ii) Response (LLM)

iii) For (i=0;i!=n;++i) {

iv) Review/adjust (Human input)

v) Response (LLM(s) - different perspectives) }

vi) Rewrite parts that failed/I am not happy with (Human input).

vii) Optionally repeat from iii.

This tends to get me somewhere I want. For code at least. I made a document with it recently (fairly concise) and still worried I'd missed some AI slop in it. Though there is always human slop if I just made it by myself.

It's our absolute obligation to make sure "meat proxy" sticks as hard as "vibe coding" did. It's too good not to.

Weve been writing articles like this since 2023 and the problem only gets worse year-after-year.

I’m sorry AI purists but you have lost the argument. I know a guy who just started at my firm and he uses LLMs all day to submit his work product.

He’s getting ahead. Coworkers don’t seem to mind it.

It is time to come to grips with the fact that people are going to be “meat proxies” and be promoted over you unfortunately. Whining on HN hasn’t changed this dynamic and it never will.

When using AI to generate content, content of any kind - then I urge the author to evaluate the true cost for the organisation, not just for him or her. Using AI to generate 20 pages of documentation in 30 minutes looks good for you - but having 100s of people read through that slop will cost the organisation, not only in plain work hours, but also in lost trust, respect, joy of work as well a sense of purpose.

And yes, I've gotten '50 pages tech spec' docs sent my way from upper management -- as a 'help'....

When "ai slop" became a concept, it mostly came into spaces where "human slop" had already been firmly established.

Obvious ones like content scroll fodder. But also, there's slop everywhere. Even things like longform journalism or trade publications... these are often "slop." A string of platitudes and generalizations that superficially seem serious.

That is true in work culture too. A lot of people jump in on emails, superficially seem to be working... but contribute little of value. Now they do this, but with ai.

Remove meat proxy before flight

The problem is that people either don't understand how economic value is created, or want to obscure their lack of contribution to it. They see that they can make the AI do some work without revealing what work they put in. They use the AI as a shield. To genuinely and clearly provide value they have to reveal their prompt and verify its response.

I wrote about this a year ago: https://blog.gpkb.org/posts/just-send-me-the-prompt/

Sending LLM outputs to people is worse than useless, because you don't even know what they prompted with. For all you know it was "write me a paragraph justifying my opinion".

> NATS control-plane events: stream leader election / R3 quorum re-form during pod churn.

> Jesus. I had to lookup almost every word to make sense of this.

There's a lot to say about this, but I would just start with why are you using NATS if you don't understand it at all, and how did this become AI's fault?

well, I dont understand code myself, so in this case the only way I could to do verify is to test the real product

> I can talk to Claude myself. It's going to be faster and I get to control the context. I don't need a meat proxy in between.

Yes you do apparently need a meat proxy, because either 1) you didn't tell them that you already talked to claude, or 2) you didn't already talk to claude. Either way they are forced to be your meat proxy because, apparently, you never asked an AI, or didn't tell them you asked an AI, and asking an AI is now table stakes.

The copying and pasting of AI output into a chat has previously been referred to as a “slop grenade”. That’s a good phrase for it!

"The CI is failing on the main branch"

Don't be a meat proxy!

Lets not be stupid about this. Do whatever works for you, and don't let some moron set the standards for your social interactions because they got the ick one morning, just because they wrote it down and put it on a website.

This is a weak argument but I hope you develop it into a better one.

When you say "X proxy" the 'X' stands for the thing that is being proxied not the thing that is doing the proxying.

I think meat proxies are solved by ARs which is the new PR, we built a tongue-in-cheek attentionrequest.com site about it, but the general concept has been working for our team -- asking people to write questions before they send an agent-generated doc, to serve as a proxy for having read / thought about it.

The real reason people do it:

1. It's quick

2. It's not their problem anymore because they've replied.

No amount of hand-wringing is going to change that.

Recent models have had a worrying lean towards longer output. If you're going to serve up some AI slop, at least distil it to the smallest possible footprint possible.

in the same vein, I have a coworker who's code reviews have magically transformed into stopping-the-world with all these great ideas for abstraction and insane edge cases... he must have gotten really good overnight!

but seriously, when I get a review like that, my first thought is that why would I read a summarized AI output, which I just did for 5 hours while whipping Claude to stop re-writing raft from scratch.

If people say: AI said "X". It usually means you should've asked AI before going to the human.

Just my boss..

I do this often for stupid questions. It's like sending someone a lmgtfy link.

Don't waste my time by asking me a question that:

A.) I know no more about than Claude, and

B.) You could have answered yourself using Claude

All of the questions I respond to this way fulfill both A and B. Responding with "Claude said" is both more polite than most ways to respond to such questions, and a subtle hint to the other person that they should be using the LLM first for those types of questions, rather than coming to me first.