logoalt Hacker News

kuhaku22today at 10:32 AM1 replyview on HN

Not to mention these patterns didn't come out of thin air. The LLMs are statistically regurgitating language from its training set, which researchers probably tuned more towards journalistic sources like the one we're reading.


Replies

CGMthrowawaytoday at 2:44 PM

You're right of course. And it's cliche now, thanks to LLMs, so outlets like the New Yorker who pride themselves on good or interesting writing (which includes avoiding cliches) ought to find alternative rhetorical structuring.