Re:AI beats human sleuth at finding problematic images in re...
by supernova87a on Wednesday October 04, 2023 @09:14PM
(#37773818)
Why is the article's title phrased with the word "problematic" rather than "duplicated" or outright "fraudulent"? Are we trying not to offend someone's sensibilities, or not leap to conclusions (when in fact, journals have been far too reluctant to do what they should have been doing about this kind of academic dishonesty)?
[ Parent
| on HN ]
Re:AI beats human sleuth at finding problematic images in re...
by huytersd on Wednesday October 04, 2023 @09:30PM
(#37773919)
Because it’s trying to collapse multiple types of “problems” into one word in the title.
Re:AI beats human sleuth at finding problematic images in re...
by jychang on Wednesday October 04, 2023 @09:33PM
(#37773933)
And also because "creating problems" that need to be double checked, is a much lower bar than "fraudulent"; especially if accusing someone of fraud can carry slander/libel charges.
Re:AI beats human sleuth at finding problematic images in re...
by whatshisface on Wednesday October 04, 2023 @09:20PM
(#37773852)
AI can't do a full investigation and identify it as fraudulent, it can only find out when it's weird-looking.
Re:AI beats human sleuth at finding problematic images in re...
by hyperhopper on Wednesday October 04, 2023 @09:49PM
(#37774045)
I am always skeptical when I see the word "problematic". It is often used to handwave over what the actual problem is, and is very often used when the supposed problem is either trivial or a subjective judgement that many people wouldn't even deem to be a problem.
Re:AI beats human sleuth at finding problematic images in re...
by bagels on Wednesday October 04, 2023 @10:29PM
(#37774250)
Things like this are why you cannot immediately know it's fraud: a bug in xerox copiers changes numbers to other numbers due to some OCR nonsense.
http://www.dkriesel.com/en/blog/2013/0802_xerox-workcentres_...