Requests for comment/AI translations in the Greenlandic Wikipedia
This is a subpage; for more information, see the Requests for comments page.
I wrote an article about this topic some months ago in the German Kurier, which you can find here. I am the only active admin in the Greenlandic Wikipedia. When I became admin, there were around 1500 articles, most of them consisting only of a few words or completely unintelligible. I deleted most of them some years ago, so now there are only around 250 articles left, which more or less consist of some sentences, which seem to be written by Greenlanders. In the last 20 years there have only been one or two Greenlandic users in the project, even though the language has around 60,000 speakers and is the official language in Greenland. There are almost no articles written in the last five years.
I learned Greenlandic as a foreign language, I am an academic in this field and have previously worked at the Greenlandic language secretary. There they are working with language technology and machine translation, and they made the first Greenlandic machine translator, which though produces a lot of mistakes. But they are very vigilant about correct language and that there aren't produced any machine translated Greenlandic texts, which could influence and harm the Greenlandic language as an indigenous language with few speakers. With nearly no possibilities to write Greenlandic without actually learning the language, Greenlandic and the Greenlandic Wikipedia only had few problems with wrong and bad language on Wikipedia, but absolutely no growth either, since there haven't been any users in the last decade whatsoever. When the language secretary launched the first machine translator, some people began spamming the Greenlandic Wikipedia with machine translated articles, which where absolutely rubbish, making no sense at all, but since the translator translates every word it doesn't know from the lexicon, to "+???", uses of the translator have been quite obvious. Then Wikimedia launched its own AI translator, which was even worse, and this one produced completely random letter sequences, that often didn't even looked like Greenlandic. Some months ago Google Translate launched a Greenlandic AI translator and this one is quite impressive since it is able to construct complex Greenlandic sentences, which seem intelligible. Nonetheless it often makes mistakes and actually it is quite bad at semantics. In the start "I am a Greenlander" was translated to "I am a fish" and "reindeer" becomes "parrot".
Now every few days some people around the world write articles in Greenlandic using Google Translate and they seem fine, but are insufficient nonetheless. I just deleted five articles from the last few days: (1) a country named "Ælbani", which should have been "Albania", I don't know, where this meaningless spelling comes from, (2) about Romania, stating that the country lies in the south of the European Arctic, using "background" as word for "flag", (3) about a Russian town, most people never have heard of, using some sentences, which ignore Greenlandic grammar and using foreign words, making it unintelligible for native speakers, (4) about "winter", but the sentences were missing half of it, (5) about Canada, stating the country has 41 inhabitants instead of 41 million.
In bigger projects there are many users, that can spot those articles and they get deleted, but in the Greenlandic Wikipedia I am the only user, who is checking, what is written and edited, and none of the users, who "write" these "articles", cannot even comprehend, what they produce. I have connections to the Greenlandic government, and they would actually see Wikipedia as a threat for the Greenlandic language, directly counteracting official Greenlandic language policies.
This is really bad for Wikipedia's reputation and it's a problem, that I am the only person in the Wikimedia Community, who is able to check at least to some extent, if the articles are intelligible or have factual og grammatical mistakes. That said, Greenlandic actually is one of the bigger small languages, and my guess is, that there are dozens of Wikimedia projects, that haven't seen a single user, that speaks the language, for years.
In my opinion this project is an absolute fiasco: Without any users you would think, that a dead project maybe does not do anything useful, but either it would harm, but actually it corrupts the Greenlandic language and the societal and political wish for language preserving. I would actually recommend to close the project (and many others), but I know, that project closure proposals almost never get accepted, so at least I want to discuss the problem and get some comments from the global community, especially other users, that have to do with indigenous languages. --Kenneth Wehr (talk) 11:29, 25 December 2024 (UTC)