Just to give an idea how LLM nonsense is affecting #Wikipedia, perhaps a quarter of threads on ANI, the Administrators' Noticeboard for Incidents, involve LLM use—either crap articles generated by LLMs, or people using LLMs to write extremely wordy, unhelpful replies to concerns about their behaviour.
That's a huge externality right there - whatever uses AI tech has, it has enabled a DDoS attack on the hardworking volunteers trying to maintain the commons-based project the damn LLMs trained on.
