Re:33-46% of workers on MTurk used LLMs in a text production...
by HideousKojima on Wednesday June 14, 2023 @01:29PM
(#36329371)
The human content will contain the n-word or other shibboleths.
[ Parent
| on HN ]
Re:33-46% of workers on MTurk used LLMs in a text production...
by jerf on Wednesday June 14, 2023 @01:38PM
(#36329497)
The "n-word" or any arbitrary other shibboleths may guarantee you didn't use OpenAI services, but they don't have any sort of lock on LLM technology and it's trivial to ask non-controlled ones to use any you like.
Re:33-46% of workers on MTurk used LLMs in a text production...
by chaxor on Wednesday June 14, 2023 @01:33PM
(#36329433)
This also doesn't make any sense, as it would be much cheaper to use local LLMs to generate text, making it likely to see generated toxic language.
Re:33-46% of workers on MTurk used LLMs in a text production...
by sorokod on Wednesday June 14, 2023 @02:41PM
(#36330577)
AI generated text is still that even with human shibboleth decorations added
Re:33-46% of workers on MTurk used LLMs in a text production...
by George83728 on Wednesday June 14, 2023 @02:45PM
(#36330642)
Deep irony to use such a dehumanizing word as evidence of humanity.
Re:33-46% of workers on MTurk used LLMs in a text production...
by EVa5I7bHFq9mnYK on Wednesday June 14, 2023 @05:24PM
(#36332916)
Its easy to sprinkle a few n. or other words onto AI generated content.