OpenAI fires human workers for using AI to do their AI-training jobs
In a display of irony so thick you could cut it with a silicon wafer, OpenAI is purging its massive army of contractors for the cardinal sin of using ChatGPT to complete the tasks designed to teach ChatGPT.
The irony is delicious: OpenAI employs over 10,000 contractors to babysit its models, but some of these workers decided to outsource their own drudgery to the very algorithms they were hired to polish. The company officially kicked out those caught utilizing AI tools, including Grammarly, to generate or refine their feedback.
This isn't just about laziness; it's about the fear of model collapse, a delightful technical term for what happens when an AI starts eating its own digital vomit and forgets how to function properly. Contractors are tasked with reviewing ChatGPT transcripts—a project dubbed Project Lily—and their output is supposed to be the gold standard for accuracy.
Internal memos show that managers are hunting for tell-tale signs of automated slop, such as repetitive phrasing and unnatural punctuation. OpenAI has even banned the use of AI-detection software because, in a stroke of genius, they’ve deemed those tools unreliable. Meanwhile, at least one rogue contractor admitted to intentionally feeding the model garbage data just to watch the world burn. The feedback loop is officially broken.
We are watching the birth of an infinite loop of mediocrity where humans are too busy using machines to do their work, and machines are too busy learning from the result of that laziness. It is the perfect corporate ouroboros, proving that the future of work is just paying people to pretend they aren't using robots to do their jobs.
Source: 404 Media
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.