[00:43:47] I'm listening back to different versions of music ive made and going huh I wonder why I didnt choose that version to release lol [01:25:47] i have that thought a lot when im releasing photographs [01:29:26] It's like damn I almost wish I released this version [01:29:37] but then I'm like but I like the other one too XD [02:14:11] https://cdn.discordapp.com/attachments/615786602454581249/1531484889826988103/ssstik.io_chickachickaboomboomb_1785116870316.mp4?ex=6a6961f3&is=6a681073&hm=34f4a92cca187d5b69802dcc90fc5570744012de947c63218f59b0bf97a23597& [02:44:18] [1/2] https://www.youtube.com/watch?v=MRpfgo-1kNQ [02:44:18] [2/2] same energy [03:14:08] Our topic went off [08:09:06] The voice is absolutely hilarious though but it feels cringy lol [08:29:22] This is off-topic. The topic is always off. [08:29:26] 😏 [10:13:05] all channels are off-topic [12:13:04] Mr Pie from the wiki.gg server [13:10:18] not a clue, I should probably follow more and have one though [13:25:31] He's the lead of wikis on wiki.gg [13:33:59] how to form a wiki alliance like indie roblox tbh [13:35:08] is it possible to have more indie `_` wiki alliance? [13:41:40] Absolutely [13:43:58] if nothing else its extremely easy to make a wiki webring which is generally what wikis that talk about partnerships are actually doing [13:44:44] on fandom it's usually about transcluding a central template including links to member wikis on the main page [13:44:47] a true technical collaboration is harder and rarer but would be nice to see more of [13:45:03] fandom has the most corporate imaginable integration of this nature but on some level does it too [13:45:37] if some miraheze wikis decide to do so which is the proper wiki to place this template? [13:45:44] login/meta/communities [13:46:26] I think I know enough people to make a Miraheze gacha wiki alliance lol. At its most basic form it could just be a discord server of admins sharing what they've done and new wikis asking questions about how to do something. [13:46:42] it seems like documenting that might have place in the gazetteer (and might be a substantial use for the gazetteer instead of hoping for random wikis to discover it), maybe something for communities wiki [13:47:01] if you want to refer to it on your global login page by all means but it is not infrastructure to advertise such things on its own merits [13:47:14] i know three chinese animation wikis moved to miraheze in 2025 (including mine), but i don't know crats or sysops of the other two wikis at all [13:47:57] Putting a central template on communities and then using scary transclusion to put it on the other wikis' main page sounds like a good idea, though figuring out the right combination of CSS to make it work on all skins could be a huge pain. [13:48:36] could be interesting still, but yeah, citizen plus vector plus custom configs would be... whew [13:48:49] unless it could be structured as the data layer and presentation is local if that's even a thing [13:49:05] scary inclusion only copies html [13:49:17] so it's possible to style locally [13:49:28] from the central place it would be bog simple structure, on wiki styled accordingly with selectors [13:50:45] [1/2] kind of off topic but I was writing this several minutes ago and figured I'd finish, on the long term list I would like to say hello to other farms not just at the technical level our tech already does but as community outreach, which by default I think would include places like the IWA, weird gloop, wiki.gg, oasis, telepedia, other free hosting; the structure, probably a backchannel [13:50:46] [2/2] to refer crosswiki abuse, events of common interest, query 'would this be interesting for you to host' if an interesting project is unfit for one place, and stuff of that nature. but, my list is long and that's onto the long burn list unless there is other initiation first [13:51:22] that sounds exciting [13:51:36] I find it appealing but it would take some organizing for sure [13:52:32] and needs structuring so the 'civil differences' are not prominent and don't come up as the worst thing would be if reps were to quibble about how things are run, not about any of that, just finding things of common interest and having that communication even if everything day to day is different [13:52:56] would you not say the weird gloop server has become a conglomerate of wikis? from my interpretation of what you want, that server is halfway there [13:54:18] Personally, I don't have a real, "formal" alliance as in like a footer containing a list of wikis or anything like that, though I do have a network of non-Fandom wikis under my administration and have "informal" alliances with other wikis/editors, often so by talking to them as friends and just sharing info about how our wikis can prosper [13:54:48] whatever people are comfortable with and like, I am not at all prescriptive on such things [13:55:13] well what I'm talking about is at the hosting level, unless weird gloop's server has cross-hosting accomodations as well [13:55:42] Which is why I view any wiki as a potential ally, at least in sprit [13:56:05] at some point I mean to be present in notable wiki spaces across the map, but this account is dead in the water and I will have to transition to raidarr 2.0 before that happens [13:56:06] We all have the same goals of building great and prosperous wikis after all [13:57:24] And even generally, when I mean we have allies in spirit, we don't have enemies or enemy wikis, beyond just the abandoned Fandom counterparts of our wikis [13:58:15] But not like "grrrrrr let's vandalize and attack fandom", more like "build our wiki and let us surpass fandom" [13:58:36] It's not that people are "enemies", rather a system, if you know what I mean [13:58:47] I would only specifically exclude or treat as hostile projects that operate in a narrow range of fraudulent, irresponsible with user data, projects that exclusively exist to inconvenience other communities (grr lets attack fandom or we don't like this prominent wiki so we exist exclusively to sap away at them without being materially different), and so forth [13:59:05] which I can't name a living example off the top of my head at the moment anyway [13:59:41] although I did consider skywiki in its operation past the bar of irresponsible and would omit or discourage for that reason [14:00:54] Don't wanna name names ofc, but calamity mod wiki is an example of a wiki that frequently vandalizes their fandom counterpart. They're a good wiki, but are hurting themselves by fighting the Fandom one [14:01:36] I don't care about the fandom calamity wiki ofc, I just don't want the good wiki to lose the wiki/seo battle [14:01:49] that's unfortunate [14:01:54] Even though it's a meme that fandom calamity is so bad that it got its own mod lol [14:02:06] there is a certain character I would additionally expect of projects in this space and not vandalizing by practice is part of it [14:02:15] Yeah [14:02:40] As much as I hate the Fandom counterparts of forked wikis for being incredibly redundant, it's gonna hurt our SEO if we directly fought them [14:02:57] Best idea is to grow the forked wikis, never touch the Fandom ones, and move on [14:03:02] And time will help us [14:03:17] indeed [14:03:40] It also sets a bad precedent and normalises wiki vandalism, which to me is more concerning. [14:04:48] Wrong reply, I meant to reply to Raidarr's message above. [14:04:56] But the meaning is still clear. [14:08:22] they should get deletion instead of vandalism? [14:09:55] It won't work [14:10:03] Ideally yes, but Fandom has no incentive to do that. [14:10:12] The deletions will be reverted if detected [14:10:41] Whatever you do with the enemy, be smart about it [14:11:26] And the smartest thing to do is just to completely abandon it. Don't advertise the Fandom wiki, don't edit it, don't vandalize it, do nothing on it if you will. [14:11:55] Instead, advertise your forked wiki, grow it, contact all the editors you can, preferably from Discord [14:12:24] There are so many things you can do thrive even amongst these challenges. Vandalizing the old wiki is not one of them. [14:12:52] tbh one of our sysops did delete many pages on our old version but fandom didn't revert them [14:13:23] This too. I remember last year, some vandal attacked a bunch of wikis across the web, including ones on Fandom and ones that are indie. They placed NSFW images where they can, including in the site notice, like on the Vampire Survivors Wiki [14:13:23] if you have actual bodies and actual people who contribute and share and care about things as your allies then even if the numbers on paper are against you for a while, you will eventually supplant the old, lopsided, increasingly irrelevant project. an example in another domain is privacyguides.org supplanting privacytools.io [14:13:38] They also recently came back and attacked a few wikis on wiki.gg [14:14:30] Alright, fair, as long as it remains undetected, that's good. But generally, the risk is not worth it. And if they do revert vandalism and deletions, don't fight back. [14:14:46] we made editors open discussion threads requiring to delete the pages they created and edited (no other ppl involved), and an sysop checked, confirmed its all authors agreed to delete, and then approved that [14:14:55] it's public discussion on the wiki [14:15:05] idk if this made fandom unable to revert them [14:15:56] that's about as straight shot as a process can get [14:16:01] ofc. the sysop who deleted lots of pages did get removed but i got removed for adding links to the new place. maybe it's i'm the crat or what idk [14:16:25] even if fandom decides to slap you anyway you would have done nothing in the wrong that comes back to your reputation, if you have done it in a procedurally fair way [14:17:32] yes, we also had a public guideline revision discussion on fandom when we just moved to miraheze [14:17:58] many editors vote for the decision to clean away articles with less than 5 or 10 sentences [14:18:24] and after the discussion and public notice period, we performed that [14:18:44] that's another way we legally deleted many pages via community consensus [14:19:13] I mean, this works I guess [14:19:19] Or merging multiple pages into one [14:19:20] the number of articles on our fandom counterpart deduced from 2k+ to currently 178 [14:19:37] mostly early pages that some authors we can't contact [14:19:45] E.g. merging 50 individual sword pages into one large swords page [14:19:49] and with enough sentences so we can't legally delete [14:19:52] Just a random example [14:20:17] And if a wiki has less than 11 pages, inactivity bot can get the wiki closed [14:20:23] In a month [14:20:38] But again, this is all risky and probably not worth doing [14:20:39] meanwhile, our current place on miraheze already has 3k+ pages, so we compete with fandom fairly [14:21:07] and their editors even copied our pages created on miraheze back there, proving our success [14:21:29] their dmca mail denies my takedown notice email—that's another thing [14:21:55] Sekaipedia? [14:22:07] pleasant goat wiki [14:22:14] Oh [14:22:41] [1/2] https://xyy.miraheze.org/wiki/Mace [14:22:42] [2/2] https://xyy.fandom.com/wiki/Mace [14:23:16] [1/2] they refuse my mail [14:23:16] [2/2] https://cdn.discordapp.com/attachments/615786602454581249/1531668369021993074/image.png?ex=6a6a0cd4&is=6a68bb54&hm=af6c7ae11fd424cb4bc082563443f4ca4b87467550e17cb014aa69bb062832be& [14:23:24] I've had multiple people independently copy pages from Vampire Survivors Wiki onto the Fandom one, which is already incredibly abandoned and inactive lol [14:23:47] Which proves our success with that wiki [14:24:07] And Pixel Gun 3D Wiki too, though it was with one user copying hundreds of pages [14:24:40] they copy our tables made with newer templates by manual wikitable [14:25:02] that's really persistent [14:26:19] Equestripedia has nearly double the articles of the Fandom MLP wiki (6,000 vs 3,200ish) but a lot of our articles are still stubs so eh. [14:26:56] stubs matter [14:27:03] but also the fandom wiki has like 1,000 articles dedicated to just galleries and stuff so it's really like three times the amount of actual content pages since we have like 5 galleries. [14:27:30] iirc weirdgloop's guide also suggests creating articles for things that aren't so important [14:27:50] the fact you have more articles than fandom does matter [14:30:06] [1/2] https://equestripedia.org/wiki/Burger [14:30:07] [2/2] yeah we have a loto of articles that arent... super important. [14:30:44] [1/2] and fandom even allows our site-notice to be on this decision [14:30:44] [2/2] https://cdn.discordapp.com/attachments/615786602454581249/1531670244848959508/image.png?ex=6a6a0e93&is=6a68bd13&hm=48d520dbc2668bbf4084afc8da1d608bc9cbd80c424577b146ff25220cca5b5f& [14:31:11] Cooked feat.. [14:31:24] borgar [14:31:32] another thing is fandom doesn [14:31:45] [1/2] 💀 [14:31:45] [2/2] i used to do most of my editing at like 2 AM. [14:32:09] I agree with this [14:32:31] I give almost everything a page on each wiki I edit [14:32:31] so moving policies and guidelines from project namespace to help namespace naturally removes them from google results [14:33:18] i was very upset when i wrote a perfect help page on fandom but it never appeared in google result no matter how i search for it [14:33:24] page granularity is very important, especially if the goal is to compete against fandom [14:33:48] when we moved out of fandom, we in reverse made use of this [14:34:29] fandom forces certain help pages to redirect to central wiki and doesn't allow local help pages to appear in google results. that's very annoying [14:35:49] if a wiki I edit is still on fandom, and is a wiki I want to fork/expand in the future, I would hold off from page granularity, but once I get to fork that wiki, I'll take advantage of it [14:36:03] ah we also did this tbh [14:36:21] saving some articles for the initial period after moving [14:36:34] for example, ball x pit wiki on fandom only has 6 pages [14:36:46] which are fairly essential, describing and listing the more important things [14:37:06] the wiki was forked to wiki.gg, and now, some of the stuff have individual pages [14:37:21] (granted, not everything yet, but this is something we should work towards to eventually) [14:38:04] another useful tip is google `"wiki name" -miraheze -fandom`, and see which websites include links to the wiki [14:38:23] or "name" site:fandom.com [14:38:28] or site:wiki.gg/site:miraheze.org [14:38:34] i messaged some personal bloggers to ask them to update the wiki, and they're pleased to do so [14:39:17] ah tbh i mean `-site:fandom.com -site:miraheze.org` [14:39:34] or `-site:domain.fandom.com -site:domain.miraheze.org` [14:40:46] google and bing search reports which websites include a link to your wiki, but usually not all and i can find all by searching on my own [14:41:42] somehow bing's report includes more entries than google. i discovered a website linking to my wiki i didn't know before [14:45:39] ah no I was thinking of "name" site:wiki.gg to find places that mention your topic/wiki [14:57:02] ah yes that works for wikis not hosted on wiki.gg [14:57:53] i can still do `"pleasant goat wiki" site:miraheze.org -site:xyy.miraheze.org` to see what other miraheze wikis mention us [15:01:17] yeah [15:45:07] [1/2] anyone knows how to contact wmf's Design System Team? i can't find a public email or talk page of them, and i think a phabricator is not suitable, but i want to reuse some icons from codex and their license asks me to notify them [15:45:07] [2/2] https://cdn.discordapp.com/attachments/615786602454581249/1531688966615466004/image.png?ex=6a6a2003&is=6a68ce83&hm=255cd5788e0b0f977498739821faca57df5bf44c2bffb46416b417255c6dbe4f& [15:46:07] [[mw:talk:design]] is this the proper place? but seems no previous threads are doing this [15:46:07] https://www.mediawiki.org/wiki/talk:design [15:46:08] [15:46:21] if you are part of the mediawiki server, you may get a more direct response I think [15:46:46] I'm not sure if there is a broad wikimedia one, I know there is also a wikipedia one [15:46:49] I wouldn't be surprised if they've disbanded the team [15:46:55] Knowing the way WMF is going [15:47:07] wmf stuff is in flux indeed so it might not be so simple, hopefully it is [15:47:14] but tbh if it's already licensed under cc by, the term about notification is not required right? as long as i comply with cc by terms [15:47:17] I know the Codex team is gone but not sure if that's the same team [15:47:50] maybe i can also leave a message on the user talk pages of each design team members to comply with their licenses [15:47:56] but i have not seen anyone did this [15:48:20] You don't need to notify them under the requirements of the license. [15:48:22] i wonder how others comply with this? for example, citizen skin uses codex icons, how did they notify the wmf design team? [15:48:38] (and WMF cannot impose further restrictions on the license like requiring you to notify them) [15:48:55] then the requirement is confusing [15:49:14] So that's probably just a gentle request that you notify them out of interest, but there's no legal obligation under the CCBYSA license to notify them as long as you use the icons under the terms of the license. [15:49:48] [1/2] i was originally on wikimedia commons tbh, it also includes this line [15:49:49] [2/2] https://cdn.discordapp.com/attachments/615786602454581249/1531690146049560759/image.png?ex=6a6a211c&is=6a68cf9c&hm=17b02669bfd3be6ad42d8b10ff1d209c113b54f980b1c5ea4041a56c0434bc39& [15:49:56] it says it's courtesy [15:50:10] i don't want to be rude [15:51:19] I suppose that's valid, but given they've made it so difficult to find out how to notify them I wouldn't worry [15:51:35] You're not doing anything that contravenes the license by not notifying them. [15:55:20] i've also created images based on other cc by or cc by-sa images on wikimedia commons, but their pages don't have similar sentences and i simply credit the original authors. that's why i'm confused these files have this special requirements [16:33:17] https://cdn.discordapp.com/attachments/615786602454581249/1531701087751635065/e9979b029a6ad6b241f769ee9ce6f5e4.jpg?ex=6a6a2b4c&is=6a68d9cc&hm=52c12afc148393f4ae931884f28b1403c2786a0a3d5eb9e68553d1a1acedf72d& [17:12:47] [1/2] https://community.fandom.com/wiki/User_blog:Pikushi/New_Community_AI_Guidelines [17:12:48] [2/2] interesting new article from fandom [17:17:01] [1/2] <@1006806660066586704> apprently you gotta join some discord server for it [17:17:02] [2/2] https://cdn.discordapp.com/attachments/615786602454581249/1531712094926405752/B8C61FA0-B239-44F8-A7EF-AC721745AC6C.png?ex=6a6a358d&is=6a68e40d&hm=deb2db5c0129d09151e8269960c9b73d9006a166f385718de7649b0b5a185e9a& [17:17:44] https://discord.com/invite/bD8cttJMYw heres the discord sever link if you need it <@1006806660066586704> [17:22:09] Fandom is like thousands of steps backwards and usually has its steps going backwards [17:22:22] But this is the one time where they moved a couple steps forwards [17:27:25] the article seems acceptable [17:33:27] It's tricky. It gets right that inclusion of AI-generated content should be a local wiki governance / editorial decision, to allow, to ban, or whatever. [17:33:27] Textual content is tricky because there have been many cases where people who write grammatically correctly, with bullet points, and a good amount of em-dashes get flagged incorrectly as being 'AI'. [17:33:27] I would rather a more nuanced guideline that unaltered textual AI content is discouraged. People who make no attempt to format, edit for quality or brevity, etc., that's the thing to potentially discourage. [17:33:27] But otherwise, Gen AI has actually be been tremendously helpful at uplifting the non-English speaking world population in terms of elevating their voices with western liberal economic democracies [19:44:21] That's laughable. So communities can decide to ban GenAI and Fandom is fine with that but they won't turn off AI generated quick answers? 🤔 [19:45:21] Interesting. [19:45:51] that is still a thing...? [19:46:40] Yup [19:46:50] Been back and forth back and forth and they won't disable it [19:47:30] [1/2] I actually only edit 1 wiki on Fandom now and very rarely, but with all the shit they're pushing onto articles this is what it currently looks like on mobile [19:47:30] [2/2] https://cdn.discordapp.com/attachments/615786602454581249/1531749963506122852/Screenshot_20260715-202618.png?ex=6a6a58d1&is=6a690751&hm=817a927a4d8c5e36a3fc8f4a15ece7c4c6b486dfdb3db1883af088985cc7b380& [19:47:39] Notice how none of the actual article content is in frame on page load [19:47:43] not to mention the AI slop tools on some wikis too [19:47:44] Just all Fandom shit and adverts [19:47:55] I don't think this is true. Machine translation has existed before the AI fad, and they were arguably better before [19:48:01] yeah this is nothing new to me sadly.. [19:48:33] even on desktop, you can face seeing zero article content if u r logged out [19:48:37] And then I asked them to remove some of the stuff and they said: "oh actually our plan was to add more stuff but I can ask for it to be disabled" [19:48:42] Tf do you mean add MORE??? [19:48:42] We have a large amount of translators who edit our wiki, and most of us agree that since Google has decided to include Gemini in Google Translate, the quality has gone way down [19:48:57] It sounds more "natural" but it's a lot less accurate [19:49:18] ai ever trying to interpret based on the least common denominator [19:49:19] another layer of ads on top of the existing ones [19:50:08] so like if you closed one dismissable ad, there will be another one behind it [19:50:17] if your cpu on fandom is not striving for triple digits, is it even trying [19:50:24] (let's not give them any ideas) [19:51:06] But it kinda already is a thing if you count fixed ads [19:51:12] Or floating ones [19:51:51] Scumbag company [19:53:21] Never forget what they said to the publishers of Satisfactory when they sought to disable ads on that wiki [19:53:52] They must pay $5m per year for that one wiki [19:54:20] well if that's not a 'go self host' statement then what is [19:55:01] AFAIK, there is one wiki on Fandom that has no ads (ignoring the bloat like the massive lazy loaded sidebar) and that is Fate Trigger Wiki [19:55:17] No way in hell they're paying $5mil/y to disable ads [19:55:59] So essentially, they're scamming devs probably like at least a few hundred times over what the ads are actually worth [19:56:44] definitely better now [19:57:06] the machine translation of the last era is cursed [19:57:54] ah this makes sense. but google allows you to switch from advanced to legacy [19:57:54] I disagree. For context, Google incorporated Gemini in December 2025 [19:58:07] i think that disables gemini [19:58:14] Meaning this is a decent addition [19:58:56] I'm not referring to old babel translations but just pre-LLM/Gemini ones :P [19:59:22] there are occasions we need it to be more natural where accuracy is not that important [20:00:45] I'm not sure which occasions those are (and still not sure if I agree with this -- or the assumption that last year's Google Translate didn't sound natural enough) but I was replying to a specific comment about how AI is democratizing communication [20:01:00] not to mention, you can use a chat interface to ask llms to translate accurately if you need accuracy [20:01:08] instead of google translate interface [20:01:23] It is not; machine translation excited before LLMs and were arguably better [20:01:35] i still disagree [20:02:06] our wiki is unofficially related to a fansub group where i do proofreadings. i know what google was in the past [20:02:57] For me, LLMs are the definition of a worthless fad -- doing things that have been done by other things and doing them worse [20:03:26] google translation before gemini was introduced was nothing better than simply guessing when context is missing [20:04:40] Agreed [20:05:07] We had this conversation at the Weird Gloop server a bit earlier [20:05:12] Here's what I have to say [20:05:15] or i'd say, as non-native english user, it's easier for me to make fluent text more accurate and technical, but it's quite hard for me to make text accurate enough more fluent for english native speakers. others sometimes misunderstand what i meant before i could use ai-powered google translate [20:05:18] Google Translate pre-Gemini was still ran on a neural network, but the difference was that it was specifically designed to translate [20:05:19] [1/2] https://cdn.discordapp.com/attachments/615786602454581249/1531754448035123420/Screenshot_20260728_230404_Discord.jpg?ex=6a6a5cff&is=6a690b7f&hm=2acca9821a1a6b545f548430cedc86e9b86b41c6dbe89a583c548cc4e5350407& [20:05:19] [2/2] https://cdn.discordapp.com/attachments/615786602454581249/1531754448496623798/Screenshot_20260728_230418_Discord.jpg?ex=6a6a5cff&is=6a690b7f&hm=c7b6e8d944fa58a4832911f9f322f3e7faab67a86bca8e62425013fccf255fa0& [20:05:26] LLMs are designed to be chatbots [20:06:30] Basically, I believe LLMs are generally useless and largely an inconvenience on wikis [20:06:59] i don't think the recent changes in google translate is not finetuned for translation task [20:07:30] I think they're trying to fit a circle into a square hole [20:08:09] basically i taught myself javascript with the help of chatgpt, starting from 2022 [20:08:35] if there weren't llms my wikis are still in fandom's garbage dump [20:09:26] if idk what js is i would never be motivated enough to leave fandom [20:10:39] more than 80% of the template documentation are written by chatgpt too. i only do manual check and approval [20:11:04] i would never create the complicated templatedata json metadata manually without llms [20:11:57] neither using the interface nor writing raw json is easy [20:14:29] ok but writing the template documentation is half the fun. i genuinely love writing documentation. it helps a lot in learning/retaining what exactly things do [20:14:56] basically i have a prompt for writing template doc. here's a template i created a few days ago (manually, with no help of llms or whatever): [[mh:xyy:t:CV]]. i simply give the template code to chatgpt and it outputs the doc i need little modification [20:14:56] https://mh.wikipedia.org/wiki/xyy:t:CV [20:14:57] [20:15:27] if there weren't llms i'd only leave the template no doc, and editors can only imitate other usage to guess how to use the template [20:16:27] [1/2] you might find the manual process of template documentation fun and are quite within rights to opt out of ai on that basis, but others do not, and as long as the result is a value add I do not inherently have a problem with someone using ai to jumpstart their creation process for areas they find tedious as long as it is the overall result is an actual working creation. This is one e [20:16:28] [2/2] xample where there is now something that otherwise would have never been [20:17:15] like others have said this before and said it better, but LLMs really create a skill learning vacuum. even if it can do everything it claims, you miss out on actually learning to do all those things. which might not be an issue personally, but might become an overarching issue if somewhere down the line, someone needs to know how to do something and no one knows anymore [20:17:17] though one could quibble if those stories overcome what is lost by the volume of thoughtless nonsense that also exists, and the tax ai corporations take out of everyone else to provide selectively useful resources on the illusion of infinity at extreme resource cost and other costs [20:17:40] but the case for individual benefit aside from the societal net result I think is there [20:18:08] i only ask llms to do what i can also do but don't bother do [20:18:25] so i can verify if it's correct and it saves my time [20:19:22] [1/2] btw, i really don't know what vacuum is created if you don't type these stupid, repeated templatedata codes [20:19:23] [2/2] https://cdn.discordapp.com/attachments/615786602454581249/1531757986035269734/image.png?ex=6a6a604a&is=6a690eca&hm=7df7b6542f25a8a05497a66c66def8c05386bce43c8db5ed721eca339696d4cf& [20:19:31] [1/2] my bother is less the possible function of ai, and I think it will never be the case that suddenly everyone forgets how to code because there will always be those with a natural passion for things. More the gross excesses of the companies that peddle it, force it everywhere, and have made a mockery out of the copyright they have ruined countless lives in other contexts to preserve an [20:19:31] [2/2] d expand. When this process pops, it will be very ugly, and it will come at drowning a lot of things out with it [20:20:04] and when I say ai I am almost exclusively referring to llm tech in particular [20:20:09] rather than saying that i have an issue with using it as a starting point, my point was more "learning how to explain something can help you learn how to do it. but if you never try explaining, how will you ever learn (or, perhaps more importantly, how will you ever know whether you like doing it or not?) [20:20:58] also I've been referring to LLMs almost exclusively in this convo lol. of course there are uses for other types of AI, including neural networks [20:21:02] I mean frankly for some people it is a point of friction that is not worth imposing because in some cases that means positive things will simply never be made because of that friction [20:21:36] so while we have it I am positive about the good things that can arise by ai managing to lower certain frictions [20:21:38] i'm always explaining how something is done to llms so that it can do it more perfectly so i don't think that's something worth worrying [20:22:02] writing prompts is more interesting than doing them myself to me mayb [20:22:04] I'm not arguing for anything here? so I'm genuinely not sure why my argument is being taken as "we should never use AI" 😅 [20:22:17] for miraheze while I am not pleased with the function of the AI and would like to see different or better, it has made the review process actually work in a way that was increasingly impossible before [20:22:23] I've referred to LLMs almost exclusively throughout this convo [20:22:41] at least on my part I don't think there has been any suggestion of the absolute extreme position [20:22:54] largely because i do think they don't add much value personally [20:23:07] (notice I've used "i" statements throughout!) [20:23:26] i didn't even mention the word 'ai' after this message btw [20:24:05] its fairly meaningless to persist on are we or are we not advocating absolute extremes in general, I don't think anyone in this conversation is, and it is a sharing of perspectives and related experience [20:24:15] but I'm mostly just explaining my position here and why i don't think the very first comment i was responding to, that AI (or LLMs in specific) has democratized communication [20:24:23] context! 😅 [20:25:02] i think people communicated fine before LLMs. neural network translations existed before LLMs! and probably could've gotten better without them! [20:25:05] really what I gathered out of the earlier conversation was two radically different experiences with google translate alone that were basically impossible to reconcile [20:25:20] yeah, i was getting that impression too [20:25:48] i've understood your point of view. but i'd also say i do benefit a lot from the improved google translate, as a non-native speaker [20:26:13] not to mention other ways to use llms [21:34:14] Google translate helped me learn foreign languages. Which is ironic. [22:05:10] honestly I just copy and paste the json and edit manually [22:21:10] I was thinking about this: we can probably use an LLM to triage vandalism so that probable vandalism can be shown more prominently. [22:22:37] There’s been a lot of things related to AI for anti vandalism, but it’s been with ANNs before [22:22:58] Re ClueBot NG and Cynthia (which I did try to get working myself some years ago) [23:05:11] I think the main challenge is that LLMs process data at a much slower rate than some of the statistics-based solutions. As a tradeoff it potentially has more context on what's going on in the page. [23:56:17] a way to go about a curated cvt feed perhaps