(From Meta) Meta AI develops a novel dataset and model to help bring more representation to Wikipedia. Sorry, what? When did they ask the community whether we wanted their help and in what form? If we are to start using AI to generate baseline text to create WIR stubs from then I want it to be written by experienced Wikimedian volunteers, open source, and nothing to do with a for-profit company that engages in inordinate privacy violations and human rights abuses. (See also algorithmic bias.) — Bilorv (talk) 22:15, 24 April 2022 (UTC)[reply]
The paper is here. They do mention WiR a few times, but it doesn't sound like they made any attempt to get in touch. Overall, it does not seem particularly impressive to me − generative language models are good for a lot of stuff, they're very interesting, and they have a lot of potential, but wrapping them up in a soup of "AI" buzzwords and glowing generalities about "we made an AI to write articles" doesn't really convey what is going on. Essentially, their model does web searches and retrieves a bunch of information, which it kind of half-assedly stitches together into text resembling a Wikipedia article. But a lot of the information is nonsensical or incorrect (see page 7: "Overall, 68% of the information in generated sections is not present in the reference text"). Of course, basically all generative language models do stuff like this, Facebook has not done anomalously badly in this respect. But it'd be a lot more impressive if they weren't trying to pass it off like this thing was creating usable articles that didn't need human review. A lot of this stuff reminds me of when ASIMO would fall on its ass. jp×g23:33, 24 April 2022 (UTC)[reply]
Thanks for doing the deeper reading, JPxG. This is actually quite reassuring to me as it confirms that the Meta nonsense is just like their recent rename from "Facebook": a shallow facade of rebranding with no material substance behind it. They are simply exploring another avenue to piggyback off our labour at the expense of our reputation without any actual engagement in our movement—like the auto-generated Facebook pages they've had for many years that copy Wikipedia en masse with little or no attribution. Here it is wrapped in a laughable veneer of progressivism, with 32%-accurate articles made by an AI that is not actually available to Wikimedians, nor useful if made available, nor desirable even if improved based on better people that have tried and failed in this area. And this helps women how? — Bilorv (talk) 20:32, 25 April 2022 (UTC)[reply]
As for the "Wikipedian" shooting up a school, I see no reason to avoid mentioning the notable Edmund Burke School, which was the target of the attack as reported by the attacker himself, and countless reliable sources. I would not have chosen to call the suicidal gunman a "Wikipedian" because, to me, that term (which I rarely use myself), describes a person committed to Wikipedia's basic principles. Although this person's edits have been rightfully suppressed, it seems clear that they were in no way and at no time a good faith contributor to this encyclopedia. I think that "recently registered Wikipedia editor" would have been a more accurate and neutral description of the gunman. In the spirit of full disclosure, I have edited the school article to add neutral content about the shootings, and I was contacted in advance about this Signpost article, which I appreciate. Cullen328 (talk) 06:02, 25 April 2022 (UTC)[reply]
I am also concerned with the heading. Is everyone who makes several edits to Wikipedia a "Wikipedian"? I think this term should be reserved for people who identify with the project goals and have done more than a few fly by edits (not to mention - are trolls and vandals "Wikipedians" too?). --Piotr Konieczny aka Prokonsul Piotrus| reply here07:37, 29 April 2022 (UTC)[reply]
Honest question: Are Larry Sanger's statements still newsworthy? At this point, it seems that he bashes Wikipedia in some outlet every three weeks, usually repeating the same arguments over and over. IMO, it looks like Sanger is only trying to stay relevant instead of making an honest effort in improving Wikipedia or the Internet at large (not to mention some of his more dubious projects). Applodion (talk) 08:56, 25 April 2022 (UTC)[reply]
Larry Sanger walked away from the one reason anyone would ever care about his opinion and has been desperately grasping at relevance for twenty years now. XOR'easter (talk) 15:18, 25 April 2022 (UTC)[reply]
Nope. He thinks that neutrality means ignoring reliable sources to come to a "neutral" conclusion that scientific and factual concepts (eg. allopathy and the fact that Islamophobia is a problem) and non-scientific and false concepts (eg. homeopathy and Hinduphobia) are equal. If he had his way, we'd probably have a lot of misinformation about COVID vaccines and magnets (web link to a Fox News affiliate). Tube·of·Light03:08, 26 April 2022 (UTC)[reply]
Sanger's statements are notable, but he's become a bit of a broken record at this point, repeating the same talking points every few months, just in slightly different ways. X-Editor (talk) 20:34, 27 April 2022 (UTC)[reply]
Ryan Kavanaugh's complaints as described don't match well with what he actually said due to lack of context. His proof for his "fantasy world" is "They carry alternative titles, such as “Most Perfect Tutnum” and “Grand High Togneme Vicarus,” - which everyone knows are just jokes, no one takes them seriously (and yes, I have one, just for the fun of it). He also claims falsely that "The coterie of editors who gain more and more access by the number of contributions they have made to Wikipedia and the length of time they have been on the site. They are ranked on multiple levels, from “Registered User” to “Admin User.”" The only limitation to access is of course our extended confirmed protection policy. And of course it leaves out the fact that he has a grudge against us because we insist on reliable sources which has caused him problems verifying his age and other issues. His comments about paid editors are about his article again. He's claiming undisclosed paid editors are paid to dislike him. I loved his claim " You must counter by paying a more senior editor to fix the damage." Signpost, this is disappointing. Doug Wellertalk14:13, 25 April 2022 (UTC)[reply]
If I had been paid for my Wikipedia contributions every time somebody accused me of being motivated by personal dislike, I'd be quite rich. But "personal dislike" is rather by-the-by when all I'm doing is reading sources and collating what they say (which is very rarely what I think). — Bilorv (talk) 20:32, 25 April 2022 (UTC)[reply]
@Doug Weller: your complaint IMHO boils down to "a lack of context" which I partly consider justified, but mostly consider to be a misunderstanding of what the "In brief" section of this column is trying to do. We can't give all the facts complete with context for all the interesting stories we see in the press about Wikipedia. Why did I find this op-ed interesting? And why did I suspect many of our readers would find it interesting? Well - he nails down the accusations that many folks who try to get an article in WP the first time. Common accusations about Wikipedia are an "exclusive club" of arrogent admins and other editors who live in a 'fantasy realm they have created" along with paid editing (or possible bribery), and the high barrier to editing. He very accuratly *listed the usual accusations*. Now how accurate are the accusations? Well, a lot of it has to do with the people coming into Wikipedia for the first time. With their expectations being completely off, these are probably the best explanations they can come up with for what the see. They'll almost never come up with the "true reason" - that they are just not notable. A large portion of the blame for this type of common reaction to Wikipedia lies with ourselves and especially with WMF communications. We need to let these misguided folks know what to expect ahead of time. If we don't there will inevitably be a lot of folks who get pissed off about how they are treated here, There's another group. IMHO possibly typified by Kavanaugh who seem to know what to expect, but want to get a free personal advertisement As far as I can tell our admins dealt with him in the proper way. Now, if you as a reader. want to learn more about this phenomena you can't expect The Signpost to say all this in the one paragraph we have available. But you can click through to the original article, which looks pretty transparent to me. You might learn something. Smallbones(smalltalk)19:26, 25 April 2022 (UTC)[reply]
I think part of what makes the general public so perplexed by our process is that "notable" to us has a narrowly defined, sometimes rigorously applied definition. Whereas to most people it means something rather different – maybe "accomplished in his/her/their field" for professionals, or "much talked about on social media" for popular subjects. Which definitely don't meet WP:GNG alone. In other words, our quasi-academic, textual-source-based and bureaucratic process is is alien and confusing to outsiders. The consensus-based stuff with distributed responsibility just adds another layer of confusion. ☆ Bri (talk) 03:25, 26 April 2022 (UTC)[reply]
The general public are almost wholly unaware that we have any limitation in scope: this is mostly flattering, as it indicates that readers really struggle to find a topic that we don't have information about. Unfortunately, this assumption that all information is welcome on Wikipedia leads to misplaced efforts by newcomers who think that any piece of information will be appreciated. In the nicest case, this is somebody doubling the length of a plot summary of a movie that was already too long. In the most confrontational, it's somebody adding a fact about a politician and then assuming that the removal must be politically motivated. — Bilorv (talk) 15:56, 26 April 2022 (UTC)[reply]
When someone expresses a desire to be "in Wikipedia" my answer is, "Well, there are often difficulties and you appear to have one of the greater ones. Far as I can see you have, though no fault of your own, the misfortune of being a living person. For many purposes being a living person is a splendid thing, but for becoming the subject of a Wikipedia biography it's a handicap that most people cannot overcome." Two days ago, however, I found more effective the answer, "The most sure-fire ticket for becoming the subject of a Wikipedia article is an obituary in the New York Times. Not a paid death notice; a genuine obituary." Jim.henderson (talk) 16:10, 30 April 2022 (UTC)[reply]
← Back to In the media