First response after a very speedy skim: shouldn't a good paper stand or fall on its own merits? The quality of the process itself, the quality of the results, the quality of the writing, etc?
Disclosure of generated text is fine and probably good, and if used in some aspect of the research process itself then by all means, let's make that clear in the methods portion of a paper... But we don't hyper-scrutinize other aspects of a scholar's process. We assess on the quality of the work itself. Why should LLM use be different?
In a discipline that often argues in defense of iteration across different aspects of the research cycle, this seems to suggest a very flat, linear model of research in which LLMs are acceptable in certain phases but not in others, as though the phases can be neatly identified.
I could totally change my mind on all of this tomorrow, lol.
Trevor Burrows
npub1u7…fcpum
2024-11-14 00:37:55 UTC
in reply to nevent1q…umcl
Author Public Key
npub1u7vvtfplxqhwesnmgt2hnszrf6hthnt08yjpja2aqf7kmjtygc5qxfcpumSeen on
wss://relay.momostr.pinkPublished at
2024-11-14 00:37:55 UTCEvent JSON
{
"id": "0dac3cb4949d37c61793396b251b5f4482704b821cbc5dcf4f6c92dd2ee04bcf",
"pubkey": "e798c5a43f302eecc27b42d579c0434eaebbcd6f392419755d027d6dc9644628",
"created_at": 1731544675,
"kind": 1,
"tags": [
[
"p",
"1b4aaa9fe0eb934bfdc3c9c1bc607dff4db9d377cf5433092802d515bbebdb21"
],
[
"proxy",
"https://techhub.social/@NearerAndFarther/113478511829021045",
"web"
],
[
"e",
"bbc94a5bdbd956336228ebf87db6556627a456a5e313338792b7f873ebdfa69d",
"",
"root",
"1b4aaa9fe0eb934bfdc3c9c1bc607dff4db9d377cf5433092802d515bbebdb21"
],
[
"proxy",
"https://techhub.social/users/NearerAndFarther/statuses/113478511829021045",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://techhub.social/users/NearerAndFarther/statuses/113478511829021045",
"pink.momostr"
],
[
"-"
]
],
"content": "First response after a very speedy skim: shouldn't a good paper stand or fall on its own merits? The quality of the process itself, the quality of the results, the quality of the writing, etc?\n\nDisclosure of generated text is fine and probably good, and if used in some aspect of the research process itself then by all means, let's make that clear in the methods portion of a paper... But we don't hyper-scrutinize other aspects of a scholar's process. We assess on the quality of the work itself. Why should LLM use be different?\n\nIn a discipline that often argues in defense of iteration across different aspects of the research cycle, this seems to suggest a very flat, linear model of research in which LLMs are acceptable in certain phases but not in others, as though the phases can be neatly identified.\n\nI could totally change my mind on all of this tomorrow, lol.",
"sig": "7544157a0f5f242551be3377dd7cac09e5ac69f3e001829cfad51d2f7f1de475a3066464addb7e5b9e386feeee5715747211529114378294c0d5e43a49766356"
}