"Meat Proxy": Unchecked AI Responses Raise Concerns

The term "meat proxy" has emerged to describe those who relay responses generated by tools like ChatGPT or Claude without reading or verifying them. Popularized after a post by developer Niklas Guhn published on August 3, the term quickly spread across social media. Beyond vocabulary, it highlights issues in the workplace, in classrooms, and even in romantic exchanges — and reignites the call for systematic human proofreading and editing.

On apps, bots are already responding to each other

Dating services like Ditto and Winggg already offer AI-generated responses for users on dating apps. A Bumble executive mentioned the possibility of a future where AI models would communicate directly with each other. In this context, trust can erode in many social interactions, as it becomes difficult to know whether one is communicating with a person or a machine.

Where "meat proxy" comes from and how it spread

The term "meat proxy" is attributed to developer Niklas Guhn, who published a brief post on August 3 to denounce the proliferation of AI-generated responses copied verbatim on Slack and other professional channels, arguing that they add no value and that he could query Claude directly. The expression quickly circulated on social media, notably through a viral post on X. Its definition has broadened to refer to anyone who transmits text, code, or other AI-generated content without reading, understanding, or validating it, with the human merely acting as a relay. Concurrently, tech leaders refer to "meat computers" to describe humans, a vocabulary that the New York Times, in May, described as pitting people against machines and presenting the former as inefficient hardware compared to a digital superintelligence.

At work and in class, what we lose by relaying AI

In the professional world, if an automated response sufficed, one could directly address a chatbot; the request made to a colleague aims to obtain human insight and perspective on code or a project. In class, simple copy-pasting hinders learning. Transparent brainstorming fosters team cohesion and skill development while limiting the introduction of AI-generated errors and embarrassing missteps. Conversely, limiting oneself to relaying AI content can impair reading comprehension, analytical skills, and harm collaboration as well as understanding of information, including familial contexts. If Claude can, for example, draft emails, the lack of human input is considered risky and fuels the concern, raised from the outset of the term, about a dilution of human opinions and skills.

Proper use of AI: verify, edit, draw inspiration rather than copy

Political leaders, lawyers, and other professionals are regularly called out for inserting AI-generated content into speeches or briefs, resulting in fabricated quotes, prompts read aloud, and other quickly mocked errors online. These episodes underscore the necessity of human verification and editing. However, using AI as an information tool does not automatically make someone a "meat proxy," and many observers contest this conflation. Recommended best practices include critically proofreading, detecting flaws, and optimizing results, as well as using AI as a source of general ideas before writing oneself, in order to preserve one's voice and the authenticity of the exchange. In the face of often faulty models and responses deemed generic or repetitive, reflecting on AI outputs becomes essential, just as adopting new habits in an era where reposting AI content as "homework" has become commonplace and where new terms like "AI slop" or "meat proxy" resonate.