We are heading for an internet Kessler syndrome, where the destruction of human-written text will cause LLMs to train off of dirty LLM-written text, causing the further destruction of human-written text and the further degradation of LLM-written text. Eventually LLMs will be useless and human-written text will not be discoverable. I pray that the answer is that people seek out spaces which are not monetized (such as the gemini protocol) so that there's no economic incentive to waste computing resources on it.