Meat proxies are insecure people, devs argue

There’s a new word for people who mindlessly send AI-generated text to their colleagues. Developers seem to agree that these “meat proxies” are just insecure about their skills.

Whenever you hear words like salty, goated, or highkey used out of context, the word has almost always been appropriated and turned into internet slang.

One term that’s taken hold on the internet is clanker, a derogatory term for humanoid robots.

But instead of insulting robots, the latest internet term targets users who fail to question AI outputs.

What’s a meat proxy?

Users who share text from chatbots, typically in an office environment, without vetting them first, are now dubbed meat proxies.

This has been described as a person who forwards AI-generated text, code, or other outputs without reading, understanding, or vetting them.

The person (the meat) is just relaying information from an AI model to another human, who could easily use the chatbot themselves.

Being called a meat proxy is an intentional insult coined by one developer who has seemingly had enough of this information relay race.

Who coined the term?

The term first appeared in a blog post written by Niklas Gruhn earlier this month.

Gruhn found that when he asks a question on Slack, asks colleagues to leave feedback under a merge or pull request, or argues with his friends on WhatsApp, he usually receives a huge chunk of text taken directly from a chatbot.

While Gruhn acknowledges that he has done this himself in the past, he notes that it would be more efficient for the recipient of the giant AI-generated response to use the chatbot directly.

“It's going to be faster, and I get to control the context. I don't need a meat proxy in between.”

The developer states that this doesn’t add value to the conversation as AI-generated text is “verbose, frequently contains all too plausible nonsense, and is increasingly jargon dense.”

Gruhn implies that relying on AI, often framed as an easy way to gather information, may actually complicate communication between real people.

But developers have their own theories.

Developers share the same frustration

On the Hacker News forum, many other developers vented their frustrations about meat proxies.

One user said they “deal with this all day long at work” as people think that asking Claude to “spit out this 300-line response” is ingenious.

The user expects this behavior from an executive or a high-level manager who doesn’t understand the ins and outs of every department. However, the user does expect the manager to try to understand what the response means.

However, when it comes to junior or senior engineers, the user feels that their colleague is just sending over an AI-generated response for them to almost proofread.

Another developer added that their managers aren’t just generating 300 lines. Instead, higher-ups are “unironically generating thousands of lines of ‘documentation’ and are now making hundreds of people review it.”

“Complete brainrot. AI psychosis is getting seriously outrageous at this point,” the developer added.

Others believe that it's all down to insecurity or potential imposter syndrome, as meat proxies may be “mediocre engineers” who need to be validated by sycophantic chatbots.

“Previously, asking questions or not understanding something was a bit painful, but now you can hear 'you’re absolutely right!' and make progress all day, every day,” the user said.

That is, until the meat proxy needs to interact with real people with varying degrees of expertise.

So, meat proxies will “try to minimize it by hiding behind AI-generated content,” the user concludes.

Why do people become meat proxies?

There are various reasons people would outsource work to AI for debates, proposals, or code development. But it's likely that they lack confidence or they're simply lazy.

Russell Twilligear, head of AI research and development at BlogBuster, told Cybernews what he thinks of meat proxies.

"Either they are far too busy and have an AI answering for them, or they are just lazy. I hate when people respond with AI nonsense as if they actually wrote it. Also, people who do this feel smarter than they actually are. It's kind of sad."

It's also a possibility that the meat proxies we encounter online aren't even real. They may be bots used to push and agenda or engagement farm.

But, those who are real likely "don't have faith in their own response or answers, so they rely on AI."

Twilligear believes that "people who do this trust AI FAR too much" despite the evidence to support why users shouldn't wholly trust it.

"AI, even as advanced as it is today, still hallucinates, and often will spit out nonsense that doesn't even come close to the topic at hand. It says a lot about the users doing this that they are lazy and don't have enough brain power to develop their own thoughts," Twilligear concludes.