AKA LLMs will have progressively lower performance if retrained on LLM output. I guess that’s more a need than a want :D
Edit history
AKA LLMs will have progressively lower performance if retrained on LLM output.
LLMs will have lower performance if retrained on LLM output.