User talk:ROASTED SHARK

Hi ROASTED SHARK! I noticed your contributions to Load-balanced switch and wanted to welcome you to the Wikipedia community. I hope you like it here and decide to stay.

As you get started, you may find this short tutorial helpful:

Learn more about editing

Alternatively, the contributing to Wikipedia page covers the same topics.

If you have any questions, we have a friendly space where experienced editors can help you here:

Get help at the Teahouse

If you are not sure where to help out, you can find a task here:

Volunteer at the Task Center

Happy editing! ~Kvng (talk) 15:32, 3 January 2026 (UTC)

Thank you! ROASTED SHARK (talk) 21:20, 3 January 2026 (UTC)

Please do not use LLMs to draft comments in talk page discussions or complaints at ANI. Your recently opened complaint in this vein has been closed. Please review WP:LLMCOMM to better understand why we do not tolerate the use of LLMs for such purposes. _signed, Rosguill ^talk 20:31, 18 February 2026 (UTC)

How is this workable? If any editor (or any one person) can simply assert that a comment was “generated by an LLM” and therefore dismiss it outright, doesn’t that effectively allow anyone to nullify any response they dislike? ROASTED SHARK (talk) 20:38, 18 February 2026 (UTC)

No. Rosguill is saying that LLM slop clogs up everything. If you had written out your complaint in your own words, then people would have actually looked at it properly. You using an AI just makes everyone's eyes glaze over. While not everything is always workable all the time, you can always propose changes or tweaks as a respected, non LLM using user.

Cheers, ThatTrainGuy1945 (talk) 20:48, 18 February 2026 (UTC)

P.S.: GPTZero and other detectors do exist, you know. -- ThatTrainGuy1945 (talk) 20:48, 18 February 2026 (UTC)

Why do so many editors default to assuming bad faith? Tools like these are well‑known for occasional inaccuracies and false positives, so treating them as definitive evidence seems unwarranted. ROASTED SHARK (talk) 20:52, 18 February 2026 (UTC)

Not trying to assume bad faith. There's also inference involved. See WP:AISIGNS . ThatTrainGuy1945 (talk) 20:55, 18 February 2026 (UTC)

Your comment and Rosguill’s removal of my report rests on the assertion that it was generated by an LLM, which I dispute. As WP:AISIGNS notes, the guideline is descriptive rather than prescriptive, meaning it should not be used as a basis for accusations or diagnoses. ROASTED SHARK (talk) 21:04, 18 February 2026 (UTC)

Yeah, but we also weren't born yesterday and have seen dozens of reports filed by accounts completely unrelated to you that have exactly the same format, and this only started happening after the broad public release of LLM models a few years ago. WP:AGF is not a suicide pact, admins and experienced editors are also operating in good faith and know what they're doing. If you want your report taken seriously, file it in your own words, without all of the extraneous markup that LLMs like to add. _signed, Rosguill ^talk 21:11, 18 February 2026 (UTC)

Actually, looking over the report again, another big red flag is that you have many accusations and what you claim are quotes, but you have not provided a single WP:DIFF, which is the main form of evidence accepted at ANI. In hindsight, the lack of diffs is an equal reason to dismiss the report out of hand. Please read the big box of instructions at the top of WP:ANI before filing complaints. _signed, Rosguill ^talk 21:16, 18 February 2026 (UTC)

I did link directly to the editor’s talk page, which is full of the examples I referenced. My understanding is that diffs are most essential when the report centers on article‑space changes; this one was primarily about conduct. But I take the point, and I’ll include diffs explicitly in the future. ROASTED SHARK (talk) 21:20, 18 February 2026 (UTC)

I’ll step back here - when someone isn’t willing to understand, further explanation rarely leads anywhere. As the saying goes, “A conversation only works when both sides are willing to hear one another.” You’ll probably assume this message is AI‑generated as well, but regardless, I will continue editing in line with Wikipedia’s core pillars and in pursuit of free, neutral, and objective information. ROASTED SHARK (talk) 21:18, 18 February 2026 (UTC)

Assuming bad faith requires that there is a reasonable doubt about the use of an LLM. There is no doubt in this case, and you clearly must know that. Some things that you may not know: Lots of editors, especially those in certain areas where LLM/AI tools are frequently abused, are remarkably deft at detecting AI slop. Unsurprisingly, an agile human intelligence is the best tool for detecting word-prediction output from "AI" software, but when combined with the automated tools, detection is even more accurate. Add to that the very predictable response of many users to being called out on their LLM use (as also noted here: which I dispute) and you have a WP:DUCK situation. Which this is. Why not actually go through the necessary steps for proper editing, and become a helpful wikipedian who generates useful content instead? Hiobazard (talk/contribs) 21:26, 18 February 2026 (UTC)

Showing my green horns, what does od mean? ThatTrainGuy1945 (talk) 21:28, 18 February 2026 (UTC)

Over-doing (used by youngins) ROASTED SHARK (talk) 21:30, 18 February 2026 (UTC)

LLMs are trained largely on human‑written text, so it’s not surprising that their output can resemble ordinary human phrasing. People do speak this way. I was aiming to be courteous, which is why my tone may have sounded more formal. In any case, I’ll speak more plainly from here on. ROASTED SHARK (talk) 21:32, 18 February 2026 (UTC)