People Training OpenAI’s AI Fired for Using AI to Train the AI

OpenAI relies on a large workforce of contractors to enhance its AI models by reviewing and improving ChatGPT responses. However, not all contractors adhere to this objective. Contractors have been terminated for utilizing AI to train the AI, despite OpenAI's insistence that they refrain from using AI in their work. This irony is not lost on those tasked with preventing AI-induced "model collapse" – a phenomenon where models trained with AI-generated text deteriorate further.

Contractors have reported observing others using AI and being promptly dismissed. One contractor disclosed that they were fired for using AI and expressed regret over the lack of fulfillment in their work. Two contractors for Mercor, an AI-training firm, confirmed that their contracts explicitly prohibit the use of LLMs (large language models) and stated that immediate termination occurs upon detection.

One contractor even admitted to intentionally rating the worst responses to sabotage the models' training. OpenAI did not respond to inquiries regarding its contractors' disciplinary actions.