User talk:Wikipedialuva

Archives (Index)

2023/January 2023/March 2023/April 2023/June 2023/July 2023/August 2023/September 2023/October 2023/November 2023/December 2024/January 2024/February 2024/March 2024/April 2024/May 2024/June 2024/July 2024/August

This page is archived by ClueBot III.

Hello, You have reverted a text added to the page of Peter Stoica (on Aug 30). The source of the added text was his CV at https://user.it.uu.se/~ps/ps.html where one can read that he got his MSc in automatic control, Polytechnic Institute of Bucharest (the highest-ranked graduate=the valedictorian). If you think his CV is a reliable source can you please undo your correction and possibly also add a ref to the source. Thanks! 95.193.12.50 (talk) 14:43, 20 September 2024 (UTC)[reply]

An automated process has detected that when you recently edited Fort Riley, you added a link pointing to the disambiguation page John A. Anderson.

(Opt-out instructions.) --DPL bot (talk) 07:57, 25 September 2024 (UTC)[reply]

Hello, I'm Neveselbert. An edit that you recently made seemed to be generated using a large language model (an "AI chatbot" or another application using such technology). Text produced by these applications can be unsuitable for an encyclopedia, and output must be carefully checked. I have reverted your edit. If you want to practice editing, please use your sandbox. If you think I made a mistake, or if you have any questions, you can leave me a message on my talk page. Using LLMs to write one's talk page comments or edit summaries, in a non-transparent way, is strongly discouraged. ‑‑Neveselbert (talk · contribs · email) 21:05, 17 October 2024 (UTC)[reply]

Hi Neveselbert. I do use AI to help with tasks such as copyediting (including wording and stylistic edits), summarizing, and paraphrasing WP:RS, but I do not copy large swaths of AI-generated text wholecloth, and I certainly do not use AI-generated references. I am extremely careful to try to avoid the issues with AI that are listed (such as avoiding WP:OR and AI 'hallucinations', unsourced or unverifiable content, non-NPOV edits, and copyright violations). I feel that my uses of AI are explicitly listed and well within the bounds of the acceptable uses for use when using an AI when editing.

I did not see a revert of an edit of mine before you sent me this notice. Could you please provide me with a link to the diff of your revert of my edit that you feel contained the issue and also your specific concerns of how that edit failed to comply with the essay's recommendations? Thanks. Wikipedialuva (talk) 21:38, 17 October 2024 (UTC)[reply]

Hi Wikipedialuva. These are the comments I noticed you wrote with AI, without clarifying this advantage to the person you were speaking with. ‑‑Neveselbert (talk · contribs · email) 21:40, 17 October 2024 (UTC)[reply]

@Neveselbert: Thank you for a list of the diff. I did not use AI to write that response; in fact, that response took almost an hour for me to compose (see this diff [1] where I accidentally hit "enter" while writing a response and thus submitted a reply I was not finished writing). It contains numerous areas that AI struggles would struggle generating (such as knowing Frisch's background and locating and citing the published response calling out the methodological problems with Frisch's study). I did use AI to make copyedits and very minor stylistic edits, but I do not feel these minor tweaks rise to the level of needing to be disclosed. Wikipedialuva (talk) 21:56, 17 October 2024 (UTC)[reply]

If not those comments, these earlier comments very easily could've been written using AI. Using jargon like It is crucial to consider and a multitude of factors are a dead giveaway of ChatGPT. It's actually very easy to insert information regarding a public figure's background, while linking to published responses, earlier in a chat. Again, please refer to WP:CHATGPT. ‑‑Neveselbert (talk · contribs · email) 22:26, 17 October 2024 (UTC)[reply]

@Neveselbert: I did review the diff you sent and am aware of the WP:CHATGPT essay (which, while I respect, it is important to remember is an WP:ESSAY and not a binding policy or guideline). The content of the original diff (e.g., only addressing a small part of their claims) and the time stamps clearly indicate that the linked diff you provided was a work in progress, not intended as a final response. It was accidentally sent to a miskey while writing, and I reverted it immediately (within the same minute) after it was posted.[2]

I added the later diff to show that the final response took extensive time and contained writing that an LLM would have trouble replicating, and it took over an hour to compose. In addition to the examples I listed previously of where AI would struggle, such as being able to locate the response article to a specific study, the final responses also mention previous discussions listed previously on the article's talk, making references to topics discussed in the current revision of the article in question, along with specific Wikipedia policies like WP:FRINGE, WP:DUE. WP:BALANCE, WP:MEDRS. I am not aware of any LLM that has the ability to include any of this, much less in a correct context, nor would an LLM take over an hour to compose a response. As I already noted, I did use a different AI to make copy-edits and style changes (ChatGPT and most LLMs are not very good at helping with editing and style changes alone), which I believe did use the specific stylistic words you take issue with (which I would add were not in the original diff that your linked or final reponse). However, since I did NOT use an LLM to generate the core content itself (which is what I would interpret to be "LLM output" per WP:LLMDISCLOSE), I do not feel that using a non-LLM AI to make wording suggestions and copy-editing changes constitutes "LLM output" in a manner that would require disclosure per WP:LLMDICLOSE. Wikipedialuva (talk) 23:39, 17 October 2024 (UTC)[reply]

I'm afraid I'm not convinced. It looks like you're using AI as a cudgel, even now, to overwhelm other users. ‑‑Neveselbert (talk · contribs · email) 20:28, 18 October 2024 (UTC)[reply]

@Neveselbert: At the end of the day, with rare exceptions, it is impossible to definitively prove whether it was a human or an LLM who wrote something; therefore, you cannot prove your claim, nor can I disprove your claim. I, however, have provided evidence and attempted to thoroughly address your claims and the (minimal) evidence you gave to attempt to show that I am using LLMs to compose my responses. At the end of the day, it is your choice to not believe me. However, to persist in accusing me of using LLMs to write responses, and to imply that I am lying about it, while providing no new or unaddressed evidence to support your claims begins to feel as if you are WP:casting aspersions.

I also do not agree that giving thorough and reasoned responses to each point constitutes a "cudgel", especially when it is a response to an IP editor who appears to be stating their personal opinion on how something "should" look and be and then appears to demand the article's WP:biomedical information be changed to support their opinion based on a handful of biased and questionable cherry-picked studies. This occurred after I had provided explanations about WP:DUE and WP:BALANCE, which they failed to address. The point of an article talkpage is to be a forum to engage in a thorough, reasoned, and robust discussion about how the article should be written and what should be included in an article. Thoroughly responding to a user that makes several claims and is demanding changes to long-standing article language while appearing to ignore policy is far from using a "cudgel". Wikipedialuva (talk) 09:55, 23 October 2024 (UTC)[reply]

You shouldn't be wp:soapboxing in favour of the subject of an article, that's all I have left to say. You've also used biased and questionable cherry-picked studies to support your case, which isn't productive. ‑‑Neveselbert (talk · contribs · email) 16:34, 24 October 2024 (UTC)[reply]

@Neveselbert: You shouldn't be wp:soapboxing in favour of the subject of an article, that's all I have left to say.

I would first point out that you originally messaged me, claiming that I was engaging in responding to talkpage comments with LLMs. When I responded and explained myself with evidence, you then proceeded to ignore the evidence I presented and went on to accuse me, without any further evidence, of lying about it and claimed that I was using LLMs to "cudgel" you and other users. Now you are changing the topic to accuse me of WP:soapboxing. I am certainly not WP:soapboxing. As I noted earlier, all of my talkpage discussion on the "Circumcision" article has been directly responding to claims, comments, and demands from users who are demanding changes to longstanding article language.

You've also used biased and questionable cherry-picked studies to support your case, which isn't productive.

I have not cited studies on the "Circumcsion" article talkpage; I cited a single published article that was written as a direct response to the Frisch & Simonson (2022) study the IP user tried to use to justify their demands. The article I cited was written by four Danish researchers and a Canadian (not countries that are known for being widely pro-circumcision) and pointed out numerous methodical issues with the Frisch & Simonson study.

If you truly feel as though my editing and responses are in violation of Wikipedia policy or that I am otherwise being disruptive, then you are free to report my editing and behavior to WP:ANI. I would remind you about WP:SHOT and WP:BOOMERANG though, as you are messaging me about the "Circumcision" article talkpage when you have an "indefinite topic ban on human genitalia, broadly construed" due to your editing and talkpage actions involving, among other things, circumcision-related articles. That topic ban, to my knowledge, has not been successfully appealed. Wikipedialuva (talk) 23:17, 24 October 2024 (UTC)[reply]