My hope for the future relies on a study indicating that after 5 or so generations of training data tainted with AI generated information, the LLM models collapsed.
Hopefully, after enough LLMs have been fed LLM data, we will arrive in an LLM-free future.