News for hackers. Stuff that matters.

A comment on CNET's AI Journalist Appears to Have Committed Extensive Plagiarism. Parent comment

Re:CNET's AI Journalist Appears to Have Committed Extensive ... by juujian on Tuesday January 24, 2023 @10:56AM (#34505205)
Yup. It's essentially like a very skillful writer with zero original thoughts--because where would those come from if the model tries to detect likely next words from the preceding string of words?
[ Parent | on HN ]
Re:CNET's AI Journalist Appears to Have Committed Extensive ... by Filligree on Tuesday January 24, 2023 @11:28AM (#34505813)
Noise, fed through a very clever variational decompressor.

Only rarely does that produce useful original thoughts, but I doubt that's any different from humans, really? The difference is that humans can second-guess themselves.

[ Parent | on HN ]