Hacker News new | ask | show | jobs
by astrange 251 days ago
I think the em-dashes are OpenAI's way of watermarking their outputs. Makes it easy to spot spambots on Twitter. Well, that and they all have noticeable favorite words.