As a follow-up to the previous thread bsky.app/profile/stea..., I'm going to post here some other strange results from YesNoError as I find them, or as others send them to me (if you want to be anonymous and/or get a second set of eyeballs). Numbering starts at 15 since the other thread had 14. // - ThreadSky

steamtraen.eu • 18 days ago

As a follow-up to the previous thread https://bsky.app/profile/steamtraen.eu/post/3lk2iualces2c, I'm going to post here some other strange results from YesNoError as I find them, or as others send them to me (if you want to be anonymous and/or get a second set of eyeballs). Numbering starts at 15 since the other thread had 14.

//

Reposted from Nick Brown

This article notes that "Among 40 papers flagged as having issues, he [i.e., me] found 14 false positives (for example, the model stating that a figure referred to in the text did not appear in the paper, when it did)".

For background, here are the cases that I believe to be false positives.

/0

Comments

steamtraen.eu•18 days ago

15: In https://arxiv.org/abs/2503.00832 the AI claims that the authors omitted the ! (factorial) operator, but they didn't. This is called out as both a "Math error" and a "Discrepancy error".

//

steamtraen.eu•18 days ago

16: In https://pubs.acs.org/doi/10.1021/jm8008579 the AI complains about "Repetitive 'Abstract' headings". This seems rather trivial, but more importantly it is just wrong. In the journal article there isn't even single 'Abstract' heading (is 0 the new 2?), and in the free NIH version there's the expected 1.

//

steamtraen.eu•18 days ago

17: In https://arxiv.org/abs/2501.020... the AI claims that a formula using values in the range (0, 1) in a denominator could cause "critical" divide-by-zero errors. But the values that go into that formula cannot, in practice, be exactly 0 or 1.

//

steamtraen.eu•18 days ago

18: In https://arxiv.org/abs/2503.05446v1 the AI appears to have stripped out a pair of parentheses and then complained that they are missing. ¯\_(ツ)_/¯

//

steamtraen.eu•17 days ago

19: In https://doi.org/10.1101/2023.10.27.564100 the AI has decided to perform a content analysis of a withdrawal notice 🤔 and concluded that *the withdrawal notice itself* has a "severity:critical" issue. This counts as another "Scientific integrity violation" on the YNE front page. ¯\_(ツ)_/¯

//

steamtraen.eu•17 days ago

20: In https://openalex.org/works/W2031284984, which is a critique of a previous article by Tegmark, the AI interprets the authors' pointing out of Tegmark's purported errors as "critical" errors themselves, even though it seems to agree with them. 🤪

Comments

Posting Rules

Reply