OpenAI fires contractors for using AI to check ChatGPT answers
OpenAI recently fired several contractors for using artificial intelligence tools to check ChatGPT's answers, something the company strictly forbids.
These workers were supposed to manually review responses, but instead turned to AI shortcuts, which could potentially contribute to model collapse.
This move is a bit at odds with Sam Altman's repeated statements that AI will take jobs.
OpenAI contractors must ensure ChatGPT accuracy
Contractors at OpenAI are expected to make sure ChatGPT gives accurate, correct replies, without help from other AIs or unreliable detection tools like GPTZero.
Even though the work can get pretty repetitive, OpenAI stresses that having real people in the loop is key.
A 2024 study published in Nature also warned that too much reliance on AI-generated content could permanently harm these systems, so sticking to strict guidelines is a must.