Re:GPT detectors are biased against non-native English writers
by freehorse on Sunday May 21, 2023 @07:50AM
(#36019979)
I think people here misunderstand the whole point, non-native speakers do not necessarily make more mistakes in general, it is about perplexity of word choices and structure. In a subsequent experiment in the paper, they used chatgpt to increase and decrease perplexity in the original non-native and native texts respectively, and the exact opposite pattern was observed.
[ Parent
| on HN ]
Re:GPT detectors are biased against non-native English writers
by darkerside on Sunday May 21, 2023 @08:04AM
(#36020051)
Perplexity?
Re:GPT detectors are biased against non-native English writers
by bigbacaloa on Sunday May 21, 2023 @08:37AM
(#36020262)
Poster was illustrating the point being made.
Re:GPT detectors are biased against non-native English writers
by lupire on Sunday May 21, 2023 @08:53AM
(#36020409)
No, poster was using a word form the linked article. Click the "download PDF" to read past the abstract.
Re:GPT detectors are biased against non-native English writers
by freehorse on Monday May 22, 2023 @03:58AM
(#36028598)
If half of commenters had skimmed through the article instead of just commenting after reading just the title or, at best, the abstract, they would have answered their comments themselves.
Re:GPT detectors are biased against non-native English writers
by olddustytrail on Sunday May 21, 2023 @01:31PM
(#36023069)
This is a measure of how well the supplied text matches what the model itself would have produced.
A low perplexity means the text isn't massively different from what it might have output itself (which might be an indicator that it was produced by a model), whereas a high perplexity suggests it's the kind of semi-random nonsense you'd expect from a student. ;)