Jump to content

Wikipedia talk:WTF? OMG! TMD TLA. ARG!

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Ironically enough

[edit]

... this page has so many shortcuts pointing to it that the list looks about fit to burst out of the essay box. æ²  2007‑08‑14t04:54z

This is a Pattern Language

[edit]

If you read a page like this, you realize that wikipedia policy/guidelines/essays actually form a Pattern Language. This is no coincidence. Earliest wikipedia policy was inherited from the Portland Pattern Repository specifically the repository's WikiWikiWeb -- Kim Bruning (talk) 19:45, 1 April 2025 (UTC)[reply]

AI Image

[edit]

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.


I feel that AI-generated images like the one on this page do not belong on Wikipedia. Yahyehoku (talk) 17:10, 3 September 2025 (UTC)[reply]

@Yahyehoku that can looked AI-generated but that does not mean it is. And AI-generated content is sometimes allowed on Wikipedia. ~Rafael (He, him) • talkguestbookprojects 22:34, 4 September 2025 (UTC)[reply]
@Rafaelthegreat Look at the comment on the image.
File:Wikipedia Alphabet Soup.png - "AI generated with post processing in Adobe Photoshop to clean up certain elements"
Wikipedia:AI guidelines - "Subject to common sense and with a number of exceptions, most images wholly generated by artificial intelligence should not be used in Wikipedia articles." Yahyehoku (talk) 20:40, 5 September 2025 (UTC)[reply]
I disagee. I think this image fits the page perfectly, is humurous, doesn't look AI, and its removal would lead to the article not having a lead image. FaviFake (talk) 19:15, 13 September 2025 (UTC)[reply]
Agree with FaviFake. If you apply common sense as per the guidelines, the image fits fine. Bishonen | tålk 19:54, 13 September 2025 (UTC).[reply]
@Juwan off-site conversations don't count as consensus as per definition everyone isnt notified or invited. Coud you explain why you dislike the image? (see my previous comment about this) FaviFake (talk) 14:44, 30 September 2025 (UTC)[reply]
@FaviFake I am aware that it does not pass consensus. for the reasoning, while this specific example is mostly harmless, AI-generated content is generally inappropriate for Wikipedia. the generated aspect of the image does look quite obvious (note the yellow "filter"), making it distracting from the point of the article. following similar consensus on other topics (mainspace, talkspace, files), editors generally don't want AI images, unless when talking about AI itself. Juwan (talk) 15:18, 30 September 2025 (UTC)[reply]
Well, we're here discussing it now, so we can at least overrule general consensus on other namespaces for this specific essay.
The yellow filter can easily be fixed, and I didn't notice it until now. But I'm not sure what you mean when you say it's distracting from the point of the article? The article is about alphabet soup and the image is literally and alphabet soup. I can't imagine what a more on-point image could look like.
Also, imo it's funny. I had a good laugh once I realized the soup was literally made of letters. FaviFake (talk) 15:26, 30 September 2025 (UTC)[reply]
I agree, I don't like it one bit Evilanimal (talk) 19:54, 24 October 2025 (UTC)[reply]
... any reason why? FaviFake (talk) 19:59, 24 October 2025 (UTC)[reply]
In their defense, the nutrition label is hidden, so they can't see all the whole grain goodness and natural ingredients used. If they did, they'd probably love a bowl for themselves. —Locke Coletcb 23:09, 23 November 2025 (UTC)[reply]
I think this image fits the essay. It is an image of the subject of the essay, namely, Wikipedia alphabet soup. It's also funny.
This page is not in mainspace; it is not an article; it is a humorous essay. We do not need to hold it to the same standards as articles. 『π』BalaM314〘talk〙 17:49, 28 October 2025 (UTC)[reply]
Precisely FaviFake (talk) 18:08, 28 October 2025 (UTC)[reply]
Huh, I never knew there was any hubbub about this image. @Yahyehoku most images wholly generated by artificial intelligence should not be used in Wikipedia articles. You are correct, for article content AI generated images are very very inappropriate (except in a vanishingly few exceptions, such as articles about an AI image). But for project-space, I don't see the harm, and in an essay straddling the line between humor and seriousness, it's good to have something to make the reader crack a smile. As to the yellow filter mentioned by @Juwan, I think you're mistaking a warm image for an artificially yellowed image. The deliberate warmness of the image is to create that homely look as hinted at in the image caption (... like grandma used to make before she was indeffed). Do I have the talent and skill to mock up a soup can label, find a setting with a kitchen or dining room like this (likely either a production studio or a heavily mocked up AirBnB/etc) and control the lighting to achieve a real result? Sure. Do I want to take the time to do all that for a funny image in project-space made on a whim? Absolutely not. —Locke Coletcb 23:09, 23 November 2025 (UTC)[reply]

New non-AI alternative

[edit]

I made a non-AI alternative at File:Wikipedia Alphabet Soup.jpg. Perhaps it could be changed to that? dot.py 09:50, 2 January 2026 (UTC)[reply]

Agreed, a real photograph is much preferable to AI slop. There's no reason to use AI when a real photograph is available. ArtemisiaGentileschiFan (talk) 14:50, 2 January 2026 (UTC)[reply]
‹The template +1 is being considered for deletion.› +1. This image is much better. Sugar Tax (talk) 14:52, 2 January 2026 (UTC)[reply]
I would support switching to the new photo immediately. 3df (talk) 18:27, 2 January 2026 (UTC)[reply]
Why? —Locke Coletcb 19:00, 2 January 2026 (UTC)[reply]
Because this is Wikipedia, not Grokipedia. ArtemisiaGentileschiFan (talk) 00:23, 3 January 2026 (UTC)[reply]
can someone please come up with a reason for changing the image other than "it's AI and I don't like AI"? FaviFake (talk) 19:11, 2 January 2026 (UTC)[reply]
Because this is Wikipedia and using an AI image (even, or especially, frivolously for a joke) contradicts this website's entire purpose and values. We don't allow AI-generated articles for a reason. This isn't Grokipedia. ArtemisiaGentileschiFan (talk) 23:20, 2 January 2026 (UTC)[reply]
It doesn't matter if I like AI or not. The reason is that it's AI and AI incenses the people here, and that image is going to continue to upset people until it is changed. Now that an editor has put in the effort to create a prop and do a photo shoot, continuing to use this AI output over an editor's work would be unacceptable.
The AI artifacts in the image, if not immediately obvious, do put it in uncanny valley for me. The can label, though apparently glossy, bears no reflection of the can's hard shadow on the left side; the shadow appears to be cast from slightly farther to our left than the can is actually positioned; the can appears to be in an area not lit by the window, whose pane does not cast a hard shadow of its own; and the image appears to have been generated while there was a bug that caused outputs to be tinted yellow. You could say this is just a golden hour shot, but the image can't seem to decide if this involves hard or soft lighting.
The human editor's photo is obviously a real photo. The prop could be improved by using a glossy instead of a matte label with a cleaner cut, and maybe some better typesetting, but it as it is now is an improvement over the current image. Other editors should be free to take another, later replacement photo of their own to try and keep continuously improving this website. 3df (talk) 01:00, 3 January 2026 (UTC)[reply]
(I didn't see that an RFC was just opened. It would be best to continue the discussion there) 3df (talk) 01:24, 3 January 2026 (UTC)[reply]
There's nothing objectively wrong with the image I created. —Locke Coletcb 16:18, 2 January 2026 (UTC)[reply]
You didn't create any image, it was AI-generated. ArtemisiaGentileschiFan (talk) 02:25, 3 January 2026 (UTC)[reply]
@ArtemisiaGentileschiFan You know nobody is forcing you to be here, and you're exhibiting WP:NOTHERE behavior. —Locke Coletcb 05:15, 3 January 2026 (UTC)[reply]
I would be careful throwing around WP:ASPERSIONS like that. ArtemisiaGentileschiFan (talk) 05:50, 3 January 2026 (UTC)[reply]
It's not an aspersion if it's true. —Locke Coletcb 06:19, 3 January 2026 (UTC)[reply]
Then go to ANI, see what they think of your behavior. ArtemisiaGentileschiFan (talk) 06:26, 3 January 2026 (UTC)[reply]
While I surely appreciate the effort, your picture is so much worse compared to the AI generated one. FaviFake (talk) 17:40, 2 January 2026 (UTC)[reply]
That's clearly not true. It's way better than the low-quality AI image. ArtemisiaGentileschiFan (talk) 23:20, 2 January 2026 (UTC)[reply]
I still have the can with the label. If needed, I could take a better photo. dot.py 23:49, 2 January 2026 (UTC)[reply]
I have just done as such (File:Wikipedia Alphabet Soup (2).jpg). dot.py 00:04, 3 January 2026 (UTC)[reply]
Copy of a comment that I made off-wiki (my gut reaction, not a well-reasoned argument). This [contrast between the two images] is such a good illustration of one of the favourite new refrains at WP:TEA/WP:AFCHELP: we want to see what you can say in your own voice, even with blemishes and imperfections, rather than perfectly proportioned [AI output]. ClaudineChionh (she/her · talk · email · global) 23:20, 2 January 2026 (UTC)[reply]
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Merge proposal

[edit]

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.


I propose merging Wikipedia:Avoid shortcut talk into this page. Their purposes are identical and I don't see a need to keep them separate. They even use the same kinds of examples! FaviFake (talk) 17:15, 22 November 2025 (UTC)[reply]

It's true that there's no rule against duplicative essays, but in this instance I haven't seen any indication that the other page is truly different. I'd like to give the author of the other essay an opportunity to explain their thinking before supporting a merge, but I am leaning that way just based on what I can see in both pages. —Locke Coletcb 23:11, 23 November 2025 (UTC)[reply]
Support merge, then we can have a separate WP:RM to sort out what the title should be (though I, too, prefer the current title). —Locke Coletcb 16:41, 29 November 2025 (UTC)[reply]
Then I oppose the deletion. We will keep both pages, one with a recognizable title and one with the in-joke gibberish, and everybody is happy. Cambalachero (talk) 14:55, 2 December 2025 (UTC)[reply]
What deletion? This is a merge discussion, not MfD.
Also, I don't understand your rationale for keeping them separate, requiring double the work to maintain them. FaviFake (talk) 16:32, 2 December 2025 (UTC)[reply]
I was not aware of this essay when I wrote the other, because it has a problem: the name. It basically goes against the very idea it preaches. I get that the intention was to be self-referential, but it fails in its basic task: if I see the name on a list, category or template, I should get the basic idea of what is the essay about. If the name is confusing or unclear, the basic and automatic reaction is to skip it and keep reading.
What about merging into this essay, but then moving to the other name? The self-referential joke in the opening should be enough. Cambalachero (talk) 04:32, 26 November 2025 (UTC)[reply]
I very much prefer the self-referential title, but it sounds like you're proposing a move rather than a merger? I think that can be discussed at RM. Do you prefer that they be merged somewhere, or remain separate? FaviFake (talk) 20:18, 28 November 2025 (UTC)[reply]
I think that Avoid Shortcut Talk provides a helpful and simple way to know what to do in using shortcuts, whereas WTF provides a sillier view that is a slightly different way to explain it. I vote oppose. FloblinTheGoblin (talk) 01:30, 13 December 2025 (UTC)[reply]
Also (sorry to keep yapping but I thought of something else) there are a large number of pages that work like this, with a serious and humorous counterpart. (E.x: Assume good faith + Assume bad faith or Verifiability + Complete bollocks, I know those are both policies and not essays but the idea still stands that the essay is just a humorous extension of an already existing page) FloblinTheGoblin (talk) 02:11, 13 December 2025 (UTC)[reply]
So are you suggesting we move the serious part to one page and leave the rest here? we currently have an almost identical serious part in both pages FaviFake (talk) 18:06, 17 December 2025 (UTC)[reply]
Oh, I didn’t notice that both pages had a serious part, you’re right though. I think that would work, to merge the serious part and tag this essay as humorous. FloblinTheGoblin (talk) 21:06, 17 December 2025 (UTC)[reply]
I don't think that's ideal. Do you really think separate pages are needed here? I worry the serious part will accidentally be rewritten again in this essay in the future, since the message obviously needs an explanation. Otherwise it's truly meaningless. FaviFake (talk) 16:00, 18 December 2025 (UTC)[reply]
Yeah, you’re right I guess the best option is merging. FloblinTheGoblin (talk) 21:09, 18 December 2025 (UTC)[reply]
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Requested move 26 December 2025

[edit]
The following is a closed discussion of a requested move. Please do not modify it. Subsequent comments should be made in a new section on the talk page. Editors desiring to contest the closing decision should consider a move review after discussing it on the closer's talk page. No further edits should be made to this discussion.

The result of the move request was: not moved. (closed by non-admin page mover) Jeffrey34555 (talk) 07:52, 2 January 2026 (UTC)[reply]


Wikipedia:WTF? OMG! TMD TLA. ARG!Wikipedia:Avoid shortcut talkWikipedia:Avoid shortcut talk – The name is a complete gibberish nonsense. It may be a "joke", but it is counterproductive as a name: nobody who reads it listed in a list, navbox or category will have the slightest clue of what the hell is the page about. The section "The message" should be enough for such jokes. Cambalachero (talk) 14:07, 26 December 2025 (UTC)[reply]

So merge and use the other name? FloblinTheGoblin (talk) 16:22, 26 December 2025 (UTC)[reply]
Oh never mind, I didn’t realize they’d already been merged. Yes, I agree. FloblinTheGoblin (talk) 16:22, 26 December 2025 (UTC)[reply]
Courtesy pings: LightlySeared, Locke Cole as participants in the previous discussion. FaviFake (talk) 20:39, 26 December 2025 (UTC)[reply]
Oppose – The gibberish is exactly the point of the essay. The name is supposed to make you say, what the hell if that? I don't understand anything. And then the essay explains the joke in the exact same way, using shortcuts. It should not be retitled.
However, I support changing all mentions of this article in all lists, navboxes, or categories and other places linking to it. People will only understand the pagename when they're on this essay, so all links to it have to be understandable. These links can simply be piped while keeping the joke intact. FaviFake (talk) 20:44, 26 December 2025 (UTC)[reply]
Never mind, since we have a redirect from avoid shortcut talk I do like wtf better. I oppose. Sorry to windsurf here, but you have a good point. FloblinTheGoblin (talk) 21:25, 26 December 2025 (UTC)[reply]
Also, it would mess up the message, which references the title. FloblinTheGoblin (talk) 21:27, 26 December 2025 (UTC)[reply]
In other words, you admit that the name is useless. And if the reason to keep it is to stick to the (bad) joke, then the essay should be tagged with {{Humorous essay}}. Cambalachero (talk) 22:41, 26 December 2025 (UTC)[reply]
Well, it’s now the main essay regarding avoiding shortcuts, and it makes an important point, just does so in a mildly silly way, à la Drop the stick and back slowly away from the horse carcass. I also feel like it’s pretty straightforward what it’s about, and the title makes it more likely to be read if stumbled upon by a newcomer. FloblinTheGoblin (talk) 23:48, 26 December 2025 (UTC)[reply]
No. I said the name is incomprehensible and can be changed everywhere except on this page, where it is funny and an integral part of the joke. FaviFake (talk) 12:29, 27 December 2025 (UTC)[reply]
Oppose, the title is the point of the essay, and like WP:DROPTHESTICK, WP:WTF is meant to evoke the feeling one has when confronted with the spaghetti soup of wikilinks that can occur in discussions. —Locke Coletcb 23:52, 26 December 2025 (UTC)[reply]
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Request for comment

[edit]
The following discussion is an archived record of a request for comment. Please do not modify it. No further edits should be made to this discussion. A summary of the conclusions reached follows.
Given this is a question of aesthetics and personal values, very little in terms of policies and guidelines have been proposed. There is clear numerical consensus for Dot.py's photos (23), over AI-generated ones (5). Since there was little interest in debating which of Dot.py's image to use, I will default to status quo to avoid further bikeshedding. Ca (talk) (non-admin closure) Edit: after reading the relevant ANI thread, I'd like to comment on potential canvassing. From my search (disclosing I am in the relevant Discord server, though did not engage in the off-wiki discussion), eight RfC participants are in the server, and out of those three directly participated in the off-wiki discussion. Still, Even when discounting those, there is clear numerical advantage for dot.py's images (15-5).

I'm starting an RFC since there seems to be a disagreement whether or not this page should use AI-generated content. The current debate is if we should use File:Wikipedia Alphabet Soup.png (AI-generated) or File:Wikipedia Alphabet Soup.jpg/File:Wikipedia Alphabet Soup (2).jpg (real photographs). Please vote either:

  • Support using AI-generated content instead of real photos
  • Support using real photos over AI-generated content

Please also explain why you support your choice. ArtemisiaGentileschiFan (talk) 00:22, 3 January 2026 (UTC)[reply]

Discussed images
Original AI image
First real photograph
Second real photograph
HTGS's suggested image
(As the creator of the real images) Support using real photos over AI-generated content. dot.py 00:25, 3 January 2026 (UTC)[reply]
I should probably explain my reasoning:
  1. Ever since I first saw this page, I've absolutely despised the AI image.
  2. The AI image has the AI "piss filter".
  3. Per ClaudineChionh: [...] we want to see what you can say in your own voice, even with blemishes and imperfections, rather than perfectly proportioned [AI output].
  4. Per ArtemisiaGentileschiFan: This isn't Grokipedia.
dot.py 00:34, 3 January 2026 (UTC)[reply]
Also per ArtemisiaGentileschiFan quoting Wikipedia's AI image guidelines: Most images wholly generated by AI should not be used in mainspace and that Community members have largely rejected making exceptions merely because an image lacks obvious errors, or because no free non-AI-generated images are available. dot.py 00:50, 3 January 2026 (UTC)[reply]
@Dot.py This is project-space, not mainspace. WP:AIIMAGES is irrelevant here. —Locke Coletcb 05:17, 3 January 2026 (UTC)[reply]
  1. Cool story bro.
  2. I explained this above, it's called color temperature, and the goal was for a warm image, not a cool image.
  3. This isn't an article or a discussion where we want to see what someone thinks, this is an image used on an essay that is meant to convey a humorous idea.
  4. Water is wet, in case you guys didn't know.
Locke Coletcb 05:20, 3 January 2026 (UTC)[reply]
For your third point, if it's just an image used on an essay that is meant to convey a humorous idea, then what do you lose from changing it to an image supported by almost everyone else in this discussion? dot.py 06:07, 3 January 2026 (UTC)[reply]
(As RFC creator) Support using real photos over AI-generated content per Wikipedia's AI image guidelines, which state that Most images wholly generated by AI should not be used in mainspace and that Community members have largely rejected making exceptions merely because an image lacks obvious errors, or because no free non-AI-generated images are available. ArtemisiaGentileschiFan (talk) 00:30, 3 January 2026 (UTC)[reply]
It's stunning that you can quote something and not understand it. —Locke Coletcb 05:24, 3 January 2026 (UTC)[reply]
Support using real photos over AI-generated content I brought this up earlier (and it does seem silly for a humorous essay), but we should try to avoid clearly AI-generated content. Much of our editor base is against AI as well, so there's no reason to make an exception here since editors are the ones this page is intended for. HurricaneZetaC 00:47, 3 January 2026 (UTC)[reply]
Can you provide a WP:PAG that disallowes AI image use in project-space? —Locke Coletcb 05:34, 3 January 2026 (UTC)[reply]
  • Comment Frankly, this is completely and utterly bizarre to me. This is not even a matter of an image being removed with no replacement. To me, it is nigh objective that the image without AI is in no way worse than the one that does use AI, so it becomes a question of why the AI-free image should be used. The image that uses AI looks, frankly, really poor. I wouldn't even say it looks good for an AI image. Like Artemisia noted, it has the piss filter, which is generally considered a flaw. What is wrong with the non-AI image, @Locke Cole:? Why should it be used over an image that will be the source of controversy from many more people that the non-AI image would not be? - Cukie Gherkin (talk) 00:56, 3 January 2026 (UTC)[reply]
  • Thanks @ArtemisiaGentileschiFan, I thought we might be getting into RfC territory here. I support dot.py's photo, and real photos in general, per my comment quoted above, and thank dot.py for their initiative. ClaudineChionh (she/her · talk · email · global) 01:10, 3 January 2026 (UTC)[reply]
    Comment: I see that Locke Cole (and possibly others) interpreted my previous comment as implying that canvassing had taken place off-wiki. The conversation I was alluding to took place on Discord and consisted of mostly technical/image-related feedback on earlier versions of dot.py's image. At no point did I feel that dot.py or anyone else in that discussion was attempting to rally support for changing the image on this page. ClaudineChionh (she/her · talk · email · global) 07:06, 3 January 2026 (UTC)[reply]
  • Comment For easy access, I'll place the images here. dot.py 02:16, 3 January 2026 (UTC)[reply]
Support. This RFC existing is absurd, the only opposer is Locke Cole. I'd boldly change the image myself but I can't figure out how to without screwing up the formatting. Athanelar (talk) 17:38, 3 January 2026 (UTC)[reply]
i'd strongly advise against it until the RfC is closed. Otherwise people wouldn't know how the image looks when cropped. FaviFake (talk) 17:49, 3 January 2026 (UTC)[reply]
also, interpolating your comments inside other editors' is generally frowned upon on WP. For the record, the images below are part of @Dot.py's comment above. FaviFake (talk) 17:51, 3 January 2026 (UTC)[reply]
I hit 'reply' on the RfC and that's where it put my comment, not sure what happened there. Athanelar (talk) 18:05, 3 January 2026 (UTC)[reply]
The {{Multiple image}} template may be messing it up, I’ll move it to the top of the RFC. dot.py 18:11, 3 January 2026 (UTC)[reply]
Comment for everyone's info, I have already reverted to the real image. I went with the first one rather than the second as I think it looks a bit better (the second one looks a bit contrasty/blown out) but I have no oppositions if someone wants to change it to the second one; anything is better than AI slop. Athanelar (talk) 18:28, 3 January 2026 (UTC)[reply]
I also think I like the first one better, it feels more like the can is on a shelf with that rack thing in the background. @Dot.py: Consider post-processing the photos for color and contrast! Not that they need it. 3df (talk) 18:50, 3 January 2026 (UTC)[reply]
I think the first real photo is fine the way it is. Aaron Liu (talk) 18:54, 3 January 2026 (UTC)[reply]
Thanks! I actually did take the photo on my pantry shelf, lol. I don't know exactly how to post-process it, and as @Aaron Liu stated, I think the first real photo is fine the way it is. If anyone else wants to post-process it, then they are completely free to. dot.py 19:45, 3 January 2026 (UTC)[reply]
  • Real image, no real preference on which, but anybody who thinks an AI image is an improvement on Wikipedia on any page on which an actual factual image is available needs a {{trout}} IMHO, and the edit-warring to keep the AI image in the page by the creator of the AI image smacks of WP:OWN. - The Bushranger One ping only 18:37, 3 January 2026 (UTC)[reply]
    But this isn’t a case of a factual image. The image is merely illustrative; it would have the same factual value if it were drawn with crayons. — HTGS (talk) 19:34, 3 January 2026 (UTC)[reply]
    I now wonder how editors would respond to an AI-generated can label that was printed out and then photographed, or an AI model printed on a 3D printer. It's real enough to do damage if you drop it on your toes, but is it still fake? WhatamIdoing (talk) 20:16, 3 January 2026 (UTC)[reply]
    I wonder how many angels can dance on the head of a pin? Athanelar (talk) 20:18, 3 January 2026 (UTC)[reply]
  • Real image preferably the first. The second one has weird dimensions, has the can glaringly askew/slanted and off-center, and I find the colors off-putting especially at the top. The AI image is horribly lit and the shadows make everything unreadable if the small size even at thumbnail size didn't make it unreadable enough. No idea why people think the second real image is better. Aaron Liu (talk) 18:41, 3 January 2026 (UTC)[reply]
    Also, "The Free Alphabet Soup". Who can resists that‽ Aaron Liu (talk) 02:36, 4 January 2026 (UTC)[reply]
    Unrelated, but I love your use of the interrobang dot.py (alt) 20:29, 5 January 2026 (UTC)[reply]
  • Prefer this real image to this AI image. For very different reasons, neither of these images are ideal, but they are (imo) equivalent in quality for the encyclopedia. AI images are fine by me in this sort of context—where they are used as an illustration of a non-factual subject and are clear and error free—but where an AI image is equivalent in quality to a photograph or human-drawn image, the non-AI image should be preferred. Of course I believe either image could be improved fairly easily, so this RFC should have been conducted with better images, but per my ceteris paribus preference for the human made photograph, I would prefer to see that image improved first. — HTGS (talk) 19:49, 3 January 2026 (UTC)[reply]
    AI generated can
    A printable label for a can—probably needs editing to fit well
    It’s probably too late in the game to amend the RfC with new options, but I would prefer a high-quality AI image (right) to a low-quality human photo. I also include the slightly absurd option of printing the AI label and using it to create a human photo with a real can.
    (If only someone with some rudimentary design experience had access to a high quality printer and a photo studio… oh wait, that’s me! It’s a shame I’m on holiday still, and can’t be bothered with a couple of hours of work for an image so unimportant as this.) — HTGS (talk) 00:14, 4 January 2026 (UTC)[reply]
Comment If anyone has a proper label printer and/or thinks that they could take a better photo than the two that I made, I would be willing to share the image of the label that I made. dot.py 19:53, 3 January 2026 (UTC)[reply]
  • Real image First, on ethical grounds: the standard you walk past is the standard you accept, and giving a winking acceptance to slop here is giving a winking endorsement to a technology that starts with eroding the scientific literature and the public trust in news media and only gets a whole lot worse from there. And, purely aesthetically speaking, don't piss on the encyclopedia, in article space or otherwise. Stepwise Continuous Dysfunction (talk) 22:12, 3 January 2026 (UTC)[reply]
    And even setting aside the image sources, the first real pic wins on composition and style. The shadows in the AI-generated image go the wrong way and impede legibility, and it is overall too slick to go with any joke in the "just like Grandma used to make" vein. That calls for a picture which looks a little scuffed up, like it was found in an old cookbook or spotted in a family photo album. I'd go with the first real photo over the second, which is a bit contrasty/blown out as noted above. Stepwise Continuous Dysfunction (talk) 22:33, 3 January 2026 (UTC)[reply]
    The "high-quality AI image" is worse in the slickness regard. The vegetables do capture the look of AI-generated food pictures, in that they appear to have been grown at the bottom of the uncanny valley (I don't trust that carrot; it is somehow part chestburster). The acronyms look less crowded in the bowl than in either of the real photos, which also goes against the desired message. Stepwise Continuous Dysfunction (talk) 19:42, 5 January 2026 (UTC)[reply]
  • Real image, and I appreciate User:Dot.py taking their time (apparently a couple of hours!) to make a can of "Wikipedia Alphabet Soup" for use on this page. Some1 (talk) 01:49, 4 January 2026 (UTC)[reply]
  • Real image since, even ignoring the AI issues, those images are just better; it's easier to see the can, fewer disruptive shadows, etc. * Pppery * it has begun... 01:59, 4 January 2026 (UTC)[reply]
  • Real image, per everyone. Randy Kryn (talk) 04:23, 4 January 2026 (UTC)[reply]
Real image, for essentially the same reasons as everyone else (and you can also add me to the list of admirers of the effort that Dot.py put into creating that soup can prop for the image), but I also think the "The Free Alphabet Soup" on the photograph is a more humorous subtitle for the can, since it riffs on Wikipedia's slogan Yellowmarkers (talk) 06:47, 4 January 2026 (UTC)[reply]
I'll also add that I prefer the first photo over the second one because the second one is slightly tilted to the left, and it's also a bit difficult to read the whole "The Free Alphabet Soup" in the second photo because of the can's curvature, but this opinion is less pronounced than the one I outlined in my previous talk page comment. Yellowmarkers (talk) 06:56, 4 January 2026 (UTC)[reply]
Real image. The physical mockup is a great piece of work, and much more in keeping with the spirit of Wikipedia. Belbury (talk) 12:14, 4 January 2026 (UTC)[reply]
I just looked at the project page and saw what I now know is the first real image. I assumed it was AI-generated because who on Earth would go to the trouble of creating a real can? I see that I was wrong now. Haven't you all got something better to do? Could I suggest watching paint dry? Phil Bridger (talk) 21:30, 4 January 2026 (UTC)[reply]
I am of the opinion that this is a bit of a silly fight, but in my opinion, keeping AI-generated content off of Wikipedia is valuable. Cukie Gherkin (talk) 21:41, 4 January 2026 (UTC)[reply]
The effort is admirable, though. XtraJovial (talkcontribs) 00:07, 5 January 2026 (UTC)[reply]
You think people who write an encyclopedia for fun wouldn't go to the trouble of mocking up a soup can? The entirety of this project depends upon people having downright strange ideas of how to have a good time. Stepwise Continuous Dysfunction (talk) 19:36, 5 January 2026 (UTC)[reply]
Real image (first one); though is it possible to take a higher quality one? jolielover♥talk 14:04, 5 January 2026 (UTC)[reply]
  • HTGS's image -- this one is the best, clearly higher quality than any of the other suggestions and who cares that it was made using AI. That somebody printed out a label, wrapped a can in it, and took a photo of that, just to avoid use of AI, is absolutely hilarious. What's next? Typing out an article with a typewriter, taking a photo of the page, and uploading that? Or should we use lead typesetting and roll some ink over it, and press it onto some newsprint? This is a new record for silliest thing I've voted on at Wikipedia. But seriously, HTGS's image is the best quality of all the options put forward so far. Levivich (talk) 16:10, 5 January 2026 (UTC)[reply]
  • Nobody is saying that AI doesn't produce high-quality images, better than life itself some say. But who cares if it's AI? Everyone else here. Reality vs. fake technology, on Wikipedia reality should win that one hands down. Randy Kryn (talk) 16:15, 5 January 2026 (UTC)[reply]
    The human-created image is no more "reality" than the computer-created image. They're all fake -- they're all images of fake soup cans. Levivich (talk) 16:18, 5 January 2026 (UTC)[reply]
Which one is more likely to have actual soup inside? That's the encyclopedic difference. Randy Kryn (talk) 16:20, 5 January 2026 (UTC)[reply]
Since when do we rank images based on their likelihood of containing soup? FaviFake (talk) 18:17, 5 January 2026 (UTC)[reply]
This is a misstatement of what Randy said. Objectively speaking, an AI image of soup cannot, as a rule, be a real can of soup. Meanwhile, the real can with the fake label can be a can of soup. It is objectively more real than the AI image. Cukie Gherkin (talk) 20:02, 5 January 2026 (UTC)[reply]
Yes, it’s much more likely that the real can is a can of soup. Despite the fact that it has had chili in it dot.py 20:14, 5 January 2026 (UTC)[reply]
I mean I think by those standards, your can is definitionally a soupish can (by virtue of its former contents), which gives it a leg up! Cukie Gherkin (talk) 20:24, 5 January 2026 (UTC)[reply]
If I understand the argument correctly, a human-generated image of a chili can covered with a fake soup can label is better than a computer-generated image of a soup can that doesn't actually exist with a fake soup can label, because the can in the human-generated image once contained something like soup, while the can in the computer-generated image never existed at all, and therefore the human-generated image is closer to reality and thus more "encyclopedic."
There is a flaw in this otherwise impeccable logic.
The flaw is that "Wikipedia Alphabet Soup" is not, in fact, a soup. Like the soup can in the computer-generated image, it does not exist in tangible form in the real world. Rather, Wikipedia Alphabet Soup refers to the use of acronyms--letters--on Wikipedia, a website. These website letters are not actually a soup in a can (trust me, I know what I'm talking about here), they are, rather, stored digitally on a computer, as 1's and 0's. Just as Wikipedia Alphabet Soup is a collection of 1's and 0's, so, too, is the computer-generated image a collection of 1's and 0's. The "soup" inside the can in the computer-generated image is 1's and 0's, just like Wikipedia Alphabet Soup is 1's and 0's, and therefore, the computer-generated image is closer to what the image is attempting to depict, and thus more encyclopedic. Levivich (talk) 20:39, 5 January 2026 (UTC)[reply]
"Thus more encyclopedic"
Ah, yes, the most important concern on this humorous essay arguing against using weird acronyms when talking with newbies. [FBDB]
All 3 images are ugly; the one of the real can is at least cute ugly instead of distorted, slimy, "Instagram add about Tai Chi Workouts for men 40+" coded. GreenLipstickLesbian💌🧸 20:43, 5 January 2026 (UTC)[reply]
There are four lights images. Levivich (talk) 20:47, 5 January 2026 (UTC)[reply]
I got that Star Trek TNG reference. I see you. —Locke Coletcb 21:05, 5 January 2026 (UTC)[reply]
@Cukie Gherkin Is this a sincere argument? I cannot imagine editors opposed to AI images in general would equally be opposed to a human-made oil painting of the same. — HTGS (talk) 20:41, 5 January 2026 (UTC)[reply]
You don't seem to understand what I said. I didn't say people would be opposed to a depiction of a can of soup. On the topic of an oil panting of a can of soup, that would also be less potentially real than a physical can. But obviously, they wouldn't be similarly opposed to an oil painting because an oil painting isn't a low-quality image that requires stealing from artists to generate. Cukie Gherkin (talk) 20:47, 5 January 2026 (UTC)[reply]
1) None of this matters, because honestly “real” isn’t important for this image, but 2) I'm now interested in understanding what you mean; in this context, is an oil painting more “real” than an AI image? Or is there something else you’re trying to say, because you’re right, I don’t seem to understand what you’re saying. — HTGS (talk) 20:55, 5 January 2026 (UTC)[reply]
HTGS's image per Levivich. This looks so much better than any other image! The quality difference between the real image and HTGS's image is immense. FaviFake (talk) 18:20, 5 January 2026 (UTC)[reply]
Yes, an oil painting is a more real thing than an AI-generated image. One was made with actual physical materials, by a person, and the other was not. - Cukie Gherkin (talk) 20:57, 5 January 2026 (UTC)[reply]
To clear some confusion, I did not directly use the design of a Campbell's soup as inspiration to make mine. I used the original AI image for that, which itself, as you said, used Campbell's cans. dot.py (alt) 21:44, 5 January 2026 (UTC)[reply]
Apologies; to be clear, my intent was more to say "if someone did use Campbell's soup as a direct inspiration, it would be different than an AI using it as the basis for something it generated". Cukie Gherkin (talk) 22:03, 5 January 2026 (UTC)[reply]
Thanks @Dot.py, and to be clear myself, no criticism was intended in pointing out that your can was inspired by existing intellectual property, whether brand-name or AI. — HTGS (talk) 23:03, 5 January 2026 (UTC)[reply]
  • Real image per nom and Chionh.  Juwan  🕊️🌈 20:20, 5 January 2026 (UTC)[reply]
  • Replace the image with text that says "Pretend this is a can of Wikipedia® Alphabet Soup" [Humor] dot.py (alt) 21:37, 5 January 2026 (UTC)[reply]
  • Comment, hopefully my last one here. Anyone is welcome to email me for a copy of the relevant Discord chats. ClaudineChionh (she/her · talk · email · global) 23:02, 5 January 2026 (UTC)[reply]
  • Support the original image (File:Wikipedia Alphabet Soup.png). The framing of this RfC, "AI-generated content" vs. "real photos" is barely short of outrageous. Many of the best images used in article-space are photos so filtered and post-processed, they represent a purely aspirational reality. Partly because no camera can reproduce what the human eye sees. Many articles also use diagrams produced using software. And finally, this is not an article. The image looks fine to me. I don't see an unpleasant yellow cast, whereas I do see a scissored line on the physically mocked-up label (which I believe was produced using rendering software?). Perhaps both are equally figments of our respective imaginations? Subjectively, I prefer the original image as a fake can, as a representation of wiki-alphabet soup, and as a picture. More importantly, there was nothing terribly wrong with Locke cole's image; unless one is absolutely opposed to the use of AI. One is no more copyright-violating than another; there is nothing artistic in the can placement, no reminiscence of an artist's known work, nothing copyrightable about the ring-pull, and an equal evocation of branded and supermarket-brand soup cans. The image is not slop, and this use of AI is not analogous either to misusing AI in article-space to picture a historical figure or to misusing AI to generate prose. It's a tiny step beyond what Dot.py did with a printer and scissors. Bumping the original image simply because it's AI is tendentious, IMO. Framing this RfC as AI vs. real, invalidating Locke cole's equal effort and greater initiative because they used a different tool—and the invective of "slop" and "piss" in associated discussions—is uncivil, and given that none of the images is clumsy or unsuitable, that's more important and the original image should be reinstated whether you agree with my personal artistic taste or not. Yngvadottir (talk) 05:39, 6 January 2026 (UTC)[reply]
    I find it a little silly to be arguing that, because a camera can't create images the same as our eyes can, that the distinction between a photo and an AI-generated image is trivial. The difference between what an eye sees and what a photo can create is a world different than what an eye sees and what an AI produces. Cukie Gherkin (talk) 08:34, 6 January 2026 (UTC)[reply]
    it's not even close to equal effort. typing words into a prompt box takes much more energy than making and printing a physical label, taping it to a can, and taking a photo from multiple angles. ltbdl (operate) 11:25, 6 January 2026 (UTC)[reply]
    [citation needed] about the energy usage because it has to be no different than asking virtual assistants in the past 10 years where multiple tasks are performed. There are warehouses that process these cans and labels on a mass scale...The Grid (talk) 14:19, 6 January 2026 (UTC)[reply]
    But the can isn't being made for this purpose, so that's all moot unless we intend on replacing eating soup with AI Cukie Gherkin (talk) 16:42, 6 January 2026 (UTC)[reply]
    well... ltbdl (jump) 01:39, 7 January 2026 (UTC)[reply]
  • This is bikesheddy and without meaningful policy, so assuming this is mostly polling preferences, then "First real photograph" as looking at it does not give the uncanny valley feeling the two on either end do, and the lighting seems better. More generally I bikeshed that the inheritance of the first image left us with not the best collection of OMG TMB TLA ARG. IAR is rarely used. GNG would be a good one. CMD (talk) 09:13, 6 January 2026 (UTC)[reply]
    I see the IAR acronym mentioned more often than GNG, I guess because I'm less active on XfDs FaviFake (talk) 09:15, 6 January 2026 (UTC)[reply]
    Interesting. I noted an IAR resurgence due to the RfC on its depreciation, but otherwise I don't see it much. I'd be curious about the results of running a few months of AN/I through some software to get shortcut counts. CMD (talk) 09:24, 6 January 2026 (UTC)[reply]
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.