One of the things LLMs will fuck up particularly greatly is the science field of corpus linguistics. Looking into large amounts of human written or spoken texts is among the best methods to make conclusions about how language works. LLMs will greatly impede creating text corpora not only by polluting all potential sources, but also by instilling even more fear in authors that any corpora compiled for any purpose might eventually be used for LLMs. I can't blame them.
#sad #AcademicChatter