For me it was the "it's not x"/"it's y" stuff and some other structures Claude is very fond of using all the time. Perhaps humans are starting to write like LLMs!
Perhaps, just perhaps, LLMs are just statistical models that literally can't create novel things, therefore any structure LLMs write was learnt from human writing?
Perhaps, just perhaps, LLMs are just statistical models that literally can't create novel things, therefore any structure LLMs write was learnt from human writing?
But who knows!