Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I thought I would share this just because I found it interesting:

https://www.pangram.com/history/a22e7372-5970-4537-b511-0416...

Pangram is an interesting company and, to me, seems to be SOTA in AI detection. This came back as “Fully AI Generated.”

It has been interesting to read about the methodology. Still not sure how I feel about it!

Either way, as other comments have suggested, important to take things with a grain of salt as always.



Here’s a fun steelman for the post being true: the author intentionally prompted an LLM to write the post in order to further anonymize themselves from their company.


If I were a whistleblower, first thing I would do is run my text through an LLM to obfuscate writing style.


Every time this happens the OP says they're not a native english speaker so they used AI to write it from their language or notes.

Wonder if the tool can detect this phenomenon?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: