[08:25:14] <wm-bb>	 <waldyrious> I would agree that that is the main difference in technical terms (i.e. how the two systems work). But the main "semantic" difference (for lack of a better word) is IMO the impact they can have in relation to people. I really liked @relgu's take in his interview for the Between the Brackets podcast, which we discussed earlier in this channel (see https://t.me/Wikifunctions/7605)  <clipped message>
[08:25:14] <wm-bb>	 <waldyrious> he says that Abstract Wikipedia fulfills precisely the niche where AI falters, i.e. it will be have most practical use for languages with little presence on the web (since the larger ones tend to have manually written Wikipedia articles anyway), whereas the current AI systems need to ingest a lot of data to build a robust statistical model, and therefore won't be able to produce  <clipped message>
[08:25:16] <wm-bb>	 <waldyrious> content in those languages as effectively.
[08:28:06] <wm-bb>	 <waldyrious> In fact, inspired by the post about the linguist who used ChatGPT to build a constructed language for fictional mud-based beings (or something along those lines), I recently tried to test ChatGPT's ability to comunicate with me in Cape Verdean Creole (a Portuguese-based creole and my native language), and while the result was decent (it understood me very well), its ability to pr <clipped message>
[08:28:07] <wm-bb>	 <waldyrious> oduce the language was nowhere near the proficiency with which it was able to write in, say, Portuguese.
[08:30:57] <wm-bb>	 <waldyrious> Of course, once Abstract Wikipedia is materialized, I'm sure these models will slurp that content right away, internalize the model for those languages, and will likely get a lot better at working with these languages...
[10:00:00] <wm-bb>	 <Sannita> Next Wikifunctions & Abstract Wikipedia Developers-Volunteers' Corner will be held *Monday 3 April, at 17:30 UTC* (https://zonestamp.toolforge.org/1680543043)
[10:00:01] <wm-bb>	 <Sannita> 
[10:00:02] <wm-bb>	 <Sannita> Link to the meeting: https://meet.jit.si/AWVolunteersCorner
[10:51:30] <wm-bb>	 <iamsabas> https://www.standard.co.uk/business/wikipedia-written-by-ai-chatbot-jimmy-wales-b1071107.html?utm_source=tldrnewsletter
[10:51:33] <wm-bb>	 <iamsabas> lolwut
[11:14:17] <wm-bb>	 <JN_Squire> Journalists still don't understand how Wikipedia works?
[11:16:37] <wm-bb>	 <AracasDragon> Apparently (re @JN_Squire: Journalists still don't understand how Wikipedia works?)
[11:17:00] <wm-bb>	 <AracasDragon> Even with sourcing an AI would not be reliable enough
[11:24:24] <wm-bb>	 <JN_Squire> And Jimmy was mostly thinking about it because... The journalist asked about it 🤦
[11:25:57] <wm-bb>	 <JN_Squire> I've read the article and it mostly states that AI write articles because it's lying. But it could be used to spot gaps where new entries could be created
[11:28:35] <wm-bb>	 <AracasDragon> “Journalism” (re @JN_Squire: And Jimmy was mostly thinking about it because... The journalist asked about it 🤦)
[11:29:21] <wm-bb>	 <AracasDragon> That's interesting, reminds me of the Recoin extension on wikidata (re @JN_Squire: I've read the article and it mostly states that AI write articles because it's lying. But it could be used to spot gaps where ne...)