Talk:Battle of Kaliabor

Missing source?

@ImperialAficionado, thank-you for the article! Could you possibly double-check reference #4? The source, Roy (1951), seems to be missing from the bibliography. If you use {{harv}}/{{sfn}} references frequently you may find this script useful – it highlights any problems with sfn references such as missing or duplicate sources. Thanks, Wham2001 (talk) 19:07, 23 March 2024 (UTC)

Thank you Wham! Imperial^[AFCND] 23:24, 23 March 2024 (UTC)

Article was deleted

A user (Koshuri) deleted this article (claims it was AI generated) and redirected it to Mir Jumla's invasion of Assam - Wikipedia - just to let @ImperialAficionado the articles author know, will also tag @Wham2001

The claim Koshuri makes is based on AI detection software - which I am unsure is valid? But in anycase I have restored the article and invite any editors to contribute to improving this article or if they wish nominating it for deletion according to Wikipedia:Deletion policy - Wikipedia

This was brought to my attention by @JunkBorax over on the Talk:Mir Jumla's invasion of Assam - Wikipedia page so cheers for the heads up; LeChatiliers Pupper (talk) 18:58, 29 March 2025 (UTC)

I am making some copyedits on this article. The sources cited, the context are all good except some peacock terms being used. Will try to fix it. Borax || (talk to Borax) 19:07, 29 March 2025 (UTC)

Redirecting is not the same as deleting, I saw this on many AfDs so to save time I directly redirected it. But yes I agree this was not the right way, I will follow the AfD process next time. Koshuri ^(グ) 19:09, 29 March 2025 (UTC)

Understood and apologies I was wrong to say you deleted it, the page history was obviously here LeChatiliers Pupper (talk) 20:48, 29 March 2025 (UTC)

@Koshuri Sultan, you should've left a message on the creator's talk page before redirecting. Else anybody wouldn't even notice such cases. Borax || (talk to Borax) 00:14, 30 March 2025 (UTC)

Since I was tagged I've given the article a closer read than I did in March last year, when I was only doing drive-by reference fixing. The sources are old but look like they might credibly discuss the topic – I don't have straightforward access to them to check. The final paragraph of "The Battle" reads like AI slop to my untrained eye and "ultimately led the battle in favor of the Mughals" doesn't make sense in context. Given the suspicion that the article is AI-generated and the fairly sparse referencing I would advise against copyediting it for style – it would be safer either to re-write from the sources (and/or, ideally, newer sources) or reinstate the redirect. Best, Wham2001 (talk) 13:44, 30 March 2025 (UTC)

Hello Wham2001. I cross checked the citations and paragraphs. The AI forked parts were peacock terms and some weasel words has been used. There is a high possibility the creator might have used the AI tools for further fabricating their English skills. I tried to remove most of the parts where AI likely exaggerated. Thankfully it didn't put anything additional by its own. Borax || (talk to Borax) 13:52, 30 March 2025 (UTC)

OK – that's good to hear! Thank-you for checking. Best, Wham2001 (talk) 14:14, 30 March 2025 (UTC)

My own analysis of the article for AI writing;

"We are uncertain about this document. If we had to classify it, it would likely be considered human"

Was the analysis GPTzero gave me, with 50% certainty that it was written by a human, and the remaining 1/2 probability split between AI and mixed. LeChatiliers Pupper (talk) 19:14, 29 March 2025 (UTC)

I don't know how much part of it you scanned but by scanning them seperately I got:

• The lead, #Background and #The Ahom forces is detected as 98% AI Generated

• Rest of #Preparations and initial conflicts as 78% AI Generated

• #Battle as 96% AI Generated

Other parts of this were copied from the main article to which it was redirected. Koshuri ^(グ) 19:51, 29 March 2025 (UTC)

I got different values when I looked at lead/background

I didnt consider the rest of the article separately just chucked it all in you may be right there is some AI writing - even a significant part of the new material might be however the sources check out. It could be a human who has AI like tendencies in writing or a human who found sources and used an AI to join material together - that would be allowed under wiki's policies

> Specifically, asking an LLM to "write a Wikipedia article" can sometimes cause the output to be outright fabrication, complete with fictitious references. It may be biased, may libel living people, or may violate copyrights. Thus, all text generated by LLMs should be verified by editors before use in articles.

But as I said I checked some of the sources as has JunkBorax they appear to check out, if you find sources that are fabrications or that dont support the material - please delete the content it might be AI fabrication but simply being AI like isnt much of a reason to junk the whole page and redirect it to a much smaller article that doesnt treat the battle at all. LeChatiliers Pupper (talk) 20:00, 29 March 2025 (UTC)

@Asilvering, how would you analyse the AI writing of the article? I've made a copyedit by replacing some peacock terms and suspecious sentances. Borax || (talk to Borax) 07:44, 30 March 2025 (UTC)

Looking at the revision before you made the copyedit, I'd say that it doesn't look like AI slop to me at a quick glance. But there are some parts that twig my AI senses - the final paragraph of the "battle" section, for example. So my guess would be that this is AI-assisted writing, using Grammarly or something like that, if there's AI involved. Or, possibly, AI was used to paraphrase sections of the sources in an attempt to avoid plagiarism, but this was done source-by-source rather than as a fully AI generated article. It's also worth pointing out that authentic, human-written Indian English may be more likely to set off AI detectors than other varieties of English. So I don't see grounds at present for concluding that this is AI slop that needs to be redirected, but it's wonky enough that I think the references should be really carefully checked.

@Koshuri Sultan, what did you use to get those results? I ran the "battle" section through gptzero and got only 56%, nowhere close to 96%. And which parts are copied from other articles? That's a more significant problem, since the copying isn't attributed. -- asilvering (talk) 15:29, 30 March 2025 (UTC)

@Asilvering Result of the battle section (from 12:39, 21 December 2024) before borax did copyediting – result Koshuri ^(グ) 18:17, 30 March 2025 (UTC)

Interesting. I didn't take out the footnotes, which doesn't usually confuse it that badly, but evidently did this time. I've concluded that the problem is AI was used to paraphrase sections of the sources in an attempt to avoid plagiarism, but this was done source-by-source rather than as a fully AI generated article, which is why it doesn't read like AI - everything I've just checked against Sarkar 1951 is clearly close-paraphrase copyvio. So we do, indeed, need to junk this page, since I'm assuming that anything throwing up a high AI% is likely to be the same way. If anyone would go through Imperial's contributions to see if they've used Sarkar 1951 in related articles, I'd appreciate it. Otherwise we're going to have to start checking a whole bunch of sources, which is more of a pain. -- asilvering (talk) 20:20, 30 March 2025 (UTC)