Duolingo, the app to work on something every day for years and be no more skilled in that ability than if you did nothing. Now fewer people will have useless jobs which is a problem since in many ways it’s difficult to survive working a useful job.
I can handle absurd sentences like “The dog is cooking the dinner”, and actually finds them beneficial because it prevents me from guessing the whole sentence.
But this is a sign that not enough human efforts are poured into create permutation of the answers.
nope, “The dog is cooking a dinner” is that kind of absurdist sentence that works. So that I just don’t guess a human on the subject position. Or ‘eating’ for the verb
It’s because a good translation is not (always) literal.
In the German version it says taglich in hamburg. In English you would indeed put an adverb (like daily) at the end. It works the other way around but it’s not really what a native English speaker would say.
Absolutely this. I’d have argued that ‘every day’ is a more idiomatic translation than ‘daily’, and what native speakers would say, but that’s irrelevant. English tends to emphasise the end of sentences as the most important part, so all these translations are correct depending on the nuance that you intend:
Daily in Hamburg, many ships arrive (as opposed to eg. cars, or few ships)
Daily, many ships arrive in Hamburg / Many ships arrive daily in Hamburg (as opposed to eg. Bremen)
Many ships arrive in Hamburg daily (as opposed to eg. weekly)
Wouldn’t question any of those constructions as a native speaker. In fact, original responders’ example was why I gave up on Duolingo myself originally, some years ago. Translating ‘future tense’ sentences from Spanish into English or back again is always going to be a matter of opinion, since English doesn’t have the verb conjugations that Spanish does. Guessing the ‘sanctified answer’ is tedious, when a lot of the time it’s not even the most natural form of a sentence.
That’s almost exactly the problem. English uses helper words exclusively for future tense, and indeed, helper words like ‘to’ to form an infinitive. ‘Will’ is the helper word to show that something is a fact, that it is definite - grammatically, it is indicative. (The sun will rise tomorrow.) ‘Would’ is the helper word to show that something is an opinion, or dependent on something else - grammatically, it is subjunctive. (If you push that, it would fall; if it was cheaper, I would buy it.)
Spanish has both helper words for future tense (conjugations of ‘ir’, analogous to ‘going to’, often used in speech) and straight-up conjugations for future tense (doesn’t exist in English; often used in writing). It also conjugates verbs differently if they’re indicative, subjunctive, or imperative (asking or telling someone to do something). This is how Spanish manages to have fifty-odd ways to conjugate every verb, which is very confusing to English speakers who make do with three ways and helper words.
Translating a ‘future tense sentence’ for Duolingo requires you to have psychic powers about whether something is fact or opinion, which helper words are wanted, and so on, and it usually comes down to guessing between multiple ‘correct’ answers, which Duo will reject all but one of.
This is frustrating, but it has always been an issue; and usually the more you advance in a language tree the more it happens, because fewer people have found the problem and reported it. It’s a human problem that comes with not considering every possibility when creating an exercise. I’d imagine that using AI (in addition to humans) would actually help reduce cases like this, since they could be detected before users run into them.
I don’t gamble, but if I did I would bet that the AI is going to teach a lot of mistakes and maybe even be the cause of someone saying something wrong, like an insult instead of a greeting or something.
As a writer on the internet with no power to stop these companies from scraping my work, you now want to teach me using someone else’s stolen words and teach someone English using mine. Go fuck yourself.
The circle of life continues, and literacy goes down. AI cannot proofread, it merely says “these letters usually go with these”. AI screws up, people get taught shit language, they use it, it gets used as training data, rinse and repeat.
Absolutely the large language models are over glorified word predictors that get it wrong. I’d go so far as to say, they get it wrong nearly all the time.
Assuming we take the Halo lore as a matter of fact and are not projecting human faults to the Covenant: I guess not really.
As soon as monarchs or religious leaders emerge it’s game over.
And I’d guess there will always be someone more rich than the other one and be in charge of something more resulting in something like a capitalist system.
Well language is a fluid thing. If more and more people get taught shit language, the language will change to match. We have far worse problems on this planet :)
I do think these AI companies grabbing what they can without giving anything back is a problem though. In my view content creators are a bit hypocritical too though. When Google scraped the internet verbatim (viewable in google cache) they didn’t mind because it gave them discoverability. Now they suddenly do care because they don’t benefit directly. Really, the stance should have been made earlier. But I do agree it should be stopped. Or content creators compensated.
This time last year, I could still see the forum posts to related lessons when I’d get something wrong. Now, when I’m told my answer is incorrect, I have nothing to go off.
I’m trying to learn the baby steps of Korean. Being able to quickly read what I did incorrectly (and why, because usually people eould explain the grammar) was great. I hate that it’s gone, and I’m considering making Busuu my main app
Expect a lot more “white collar workers laid off due to AI” posts coming. I wonder how long it will take for a (very well resourced, those are status-y jobs) movement to form in response.
Well, that’s a bit of a salty tangent, but yeah, I guess they could take a class warfare sort of line on it. The other classical options are going full luddite, or just blaming a minority. Maybe they’ll come up with something new, because I have trouble picturing laid off creatives spouting any of these.
Right now, I think people are firmly in the denial stage. For whatever reason the thread isn’t federating properly for me, but on beehaw I can see others in here saying human exceptionalism stuff, which is kind of not in accordance with science.
I think a lot about writing a story about some sort of Enshittification Avenger. So when a reasonably good service decides to enshittify, the avenger breaks into their board’s house and beats the living shit out of them.
Disappointing, but not surprising. I know I’m not going to “learn” a language with Duolingo, but it’s been nice recognizing a few words and phrases when I hear them. But I don’t really trust that a bunch of overworked and underpaid contractors are going to catch every error using AI is going to introduce. At least there are already alternatives in this thread for me to look through.
It generally doesn’t have a high opinion of translators (note that the emojis here are inserted as path markers to help with prompt debugging - but everyting else is from the LLM):
…soon to come to your favorite corporation’s C suite’s Windows 11 desktop’s Copilot assistant for empowering the synergies of staying relevant in a high stakes market environment.
I use Duolingo for German but I’d happily switch to something else if they’re going to pull this shit. I’ll often times take things from Duolingo and run them through the Translate app on iOS to see if there are differences. It’s not ideal, but I also have no allegiance to companies.
🤖 I’m a bot that provides automatic summaries for articles:
Click here to see the summary
The popular language-learning app Duolingo cut 10 percent of its contracted translators last month amid a push to integrate generative AI into its services, multiple outlets have reported.
It’s another alarming turn in an increasingly AI-laden labor market in which company leaders continue to implement automated technology wherever they can — often, as in this case, at the cost of human jobs.
According to Bloomberg, the firings were doled out just a few weeks after Duolingo bragged in a November letter to shareholders that the company was harnessing AI to produce “new content dramatically faster.”
Duolingo also reportedly uses AI to generate some of the voices heard in various in-app language scripts and to prompt AI-generated feedback to users.
To make matters even more depressing: in a late December Reddit thread, a site user claiming to be one of the fired Duolingo translators alleged that their former team’s remaining contractors are now tasked with simply checking AI-generated text for errors.
Trusting translation AI — meanwhile pushing remaining contractors to fact-check presumably high numbers of those “dramatically faster” content outputs — may well come at the cost of such nuance, potentially flattening the learning process and rendering language robotic.
duolingo is a textbook example of a nice small startup, with great ideas that is then completely overtaken my MBAs who run it into the ground as soon as there is enough of a client base to Sell. you fucking fucks all suck.
Similar to Memrise, which was really fun when you could make your own mems using imagesearch and customise everything… And now is a rubbish duolingo clone.
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: !technology@beehaw.org
A nice place to discuss rumors, happenings, innovations, and challenges in the technology sphere. We also welcome discussions on the intersections of technology and society. If it’s technological news or discussion of technology, it probably belongs here.
Remember the overriding ethos on Beehaw: Be(e) Nice. Each user you encounter here is a person, and should be treated with kindness (even if they’re wrong, or use a Linux distro you don’t like). Personal attacks will not be tolerated.
deleted by creator
I need some more copium. Not Again :'(
Much will be lost. Language is human. Idioms and more will be missed. There is no doubt that the Duolingo product will not be as accurate.
Duolingo, the app to work on something every day for years and be no more skilled in that ability than if you did nothing. Now fewer people will have useless jobs which is a problem since in many ways it’s difficult to survive working a useful job.
I just said to someone yesterday on Mastodon that it seems as though they’re not using humans any more, because WTF is this shit?
Yeah, this is frustrating.
I can handle absurd sentences like “The dog is cooking the dinner”, and actually finds them beneficial because it prevents me from guessing the whole sentence.
But this is a sign that not enough human efforts are poured into create permutation of the answers.
I actually see a learning purpose in those ridicilous sentences.
I’ll far more likely remember the cat that works at the small hospital than if Juan does it.
Let me guess, the full sentence was: “Last night we ate the dog cooked for dinner”… /s
nope, “The dog is cooking a dinner” is that kind of absurdist sentence that works. So that I just don’t guess a human on the subject position. Or ‘eating’ for the verb
Interesting, I usually question my English skills if something like this happens!
It’s because a good translation is not (always) literal.
In the German version it says taglich in hamburg. In English you would indeed put an adverb (like daily) at the end. It works the other way around but it’s not really what a native English speaker would say.
Not true at all. OP’s construction is perfectly valid english.
Absolutely this. I’d have argued that ‘every day’ is a more idiomatic translation than ‘daily’, and what native speakers would say, but that’s irrelevant. English tends to emphasise the end of sentences as the most important part, so all these translations are correct depending on the nuance that you intend:
Wouldn’t question any of those constructions as a native speaker. In fact, original responders’ example was why I gave up on Duolingo myself originally, some years ago. Translating ‘future tense’ sentences from Spanish into English or back again is always going to be a matter of opinion, since English doesn’t have the verb conjugations that Spanish does. Guessing the ‘sanctified answer’ is tedious, when a lot of the time it’s not even the most natural form of a sentence.
Isn’t English able to disambiguate by using helper words like “will” or “would”?
What tenses can’t be translated completely?
That’s almost exactly the problem. English uses helper words exclusively for future tense, and indeed, helper words like ‘to’ to form an infinitive. ‘Will’ is the helper word to show that something is a fact, that it is definite - grammatically, it is indicative. (The sun will rise tomorrow.) ‘Would’ is the helper word to show that something is an opinion, or dependent on something else - grammatically, it is subjunctive. (If you push that, it would fall; if it was cheaper, I would buy it.)
Spanish has both helper words for future tense (conjugations of ‘ir’, analogous to ‘going to’, often used in speech) and straight-up conjugations for future tense (doesn’t exist in English; often used in writing). It also conjugates verbs differently if they’re indicative, subjunctive, or imperative (asking or telling someone to do something). This is how Spanish manages to have fifty-odd ways to conjugate every verb, which is very confusing to English speakers who make do with three ways and helper words.
Translating a ‘future tense sentence’ for Duolingo requires you to have psychic powers about whether something is fact or opinion, which helper words are wanted, and so on, and it usually comes down to guessing between multiple ‘correct’ answers, which Duo will reject all but one of.
This is frustrating, but it has always been an issue; and usually the more you advance in a language tree the more it happens, because fewer people have found the problem and reported it. It’s a human problem that comes with not considering every possibility when creating an exercise. I’d imagine that using AI (in addition to humans) would actually help reduce cases like this, since they could be detected before users run into them.
I don’t gamble, but if I did I would bet that the AI is going to teach a lot of mistakes and maybe even be the cause of someone saying something wrong, like an insult instead of a greeting or something.
Welcome to Costco, I love you.
Sayonara, Duolingo.
As a writer on the internet with no power to stop these companies from scraping my work, you now want to teach me using someone else’s stolen words and teach someone English using mine. Go fuck yourself.
The circle of life continues, and literacy goes down. AI cannot proofread, it merely says “these letters usually go with these”. AI screws up, people get taught shit language, they use it, it gets used as training data, rinse and repeat.
Absolutely the large language models are over glorified word predictors that get it wrong. I’d go so far as to say, they get it wrong nearly all the time.
The Extinction level meteor can’t come soon enough.
Time to pack it in and give some other microorganism a shot at the evolutionary big-leagues. Maybe they’ll do better.
Assuming we take the Halo lore as a matter of fact and are not projecting human faults to the Covenant: I guess not really.
As soon as monarchs or religious leaders emerge it’s game over.
And I’d guess there will always be someone more rich than the other one and be in charge of something more resulting in something like a capitalist system.
Sadly true.
Well language is a fluid thing. If more and more people get taught shit language, the language will change to match. We have far worse problems on this planet :)
I do think these AI companies grabbing what they can without giving anything back is a problem though. In my view content creators are a bit hypocritical too though. When Google scraped the internet verbatim (viewable in google cache) they didn’t mind because it gave them discoverability. Now they suddenly do care because they don’t benefit directly. Really, the stance should have been made earlier. But I do agree it should be stopped. Or content creators compensated.
This time last year, I could still see the forum posts to related lessons when I’d get something wrong. Now, when I’m told my answer is incorrect, I have nothing to go off.
I’m trying to learn the baby steps of Korean. Being able to quickly read what I did incorrectly (and why, because usually people eould explain the grammar) was great. I hate that it’s gone, and I’m considering making Busuu my main app
Expect a lot more “white collar workers laid off due to AI” posts coming. I wonder how long it will take for a (very well resourced, those are status-y jobs) movement to form in response.
The movement of hating rich people, and pretending like they’re not part of our society? I think it already exists.
Well, that’s a bit of a salty tangent, but yeah, I guess they could take a class warfare sort of line on it. The other classical options are going full luddite, or just blaming a minority. Maybe they’ll come up with something new, because I have trouble picturing laid off creatives spouting any of these.
Right now, I think people are firmly in the denial stage. For whatever reason the thread isn’t federating properly for me, but on beehaw I can see others in here saying human exceptionalism stuff, which is kind of not in accordance with science.
I think a lot about writing a story about some sort of Enshittification Avenger. So when a reasonably good service decides to enshittify, the avenger breaks into their board’s house and beats the living shit out of them.
Disappointing, but not surprising. I know I’m not going to “learn” a language with Duolingo, but it’s been nice recognizing a few words and phrases when I hear them. But I don’t really trust that a bunch of overworked and underpaid contractors are going to catch every error using AI is going to introduce. At least there are already alternatives in this thread for me to look through.
I wasn’t quite sure what to think about this, so I’ve asked my local LLM. Seems it is fine.
Holy shit it’s on the money
It generally doesn’t have a high opinion of translators (note that the emojis here are inserted as path markers to help with prompt debugging - but everyting else is from the LLM):
…soon to come to your favorite corporation’s C suite’s Windows 11 desktop’s Copilot assistant for empowering the synergies of staying relevant in a high stakes market environment.
lmao what’s wrong with your llama
I use Duolingo for German but I’d happily switch to something else if they’re going to pull this shit. I’ll often times take things from Duolingo and run them through the Translate app on iOS to see if there are differences. It’s not ideal, but I also have no allegiance to companies.
Clozemaster!
Coming from Duolingo I found it a bit overwhelming. I tried one collection, but as I already know Spanish a bit, it was too easy. Any tips?
Scroll down to the bottom and do the random collection.
🤖 I’m a bot that provides automatic summaries for articles:
Click here to see the summary
The popular language-learning app Duolingo cut 10 percent of its contracted translators last month amid a push to integrate generative AI into its services, multiple outlets have reported.
It’s another alarming turn in an increasingly AI-laden labor market in which company leaders continue to implement automated technology wherever they can — often, as in this case, at the cost of human jobs.
According to Bloomberg, the firings were doled out just a few weeks after Duolingo bragged in a November letter to shareholders that the company was harnessing AI to produce “new content dramatically faster.”
Duolingo also reportedly uses AI to generate some of the voices heard in various in-app language scripts and to prompt AI-generated feedback to users.
To make matters even more depressing: in a late December Reddit thread, a site user claiming to be one of the fired Duolingo translators alleged that their former team’s remaining contractors are now tasked with simply checking AI-generated text for errors.
Trusting translation AI — meanwhile pushing remaining contractors to fact-check presumably high numbers of those “dramatically faster” content outputs — may well come at the cost of such nuance, potentially flattening the learning process and rendering language robotic.
Saved 52% of original text.
duolingo is a textbook example of a nice small startup, with great ideas that is then completely overtaken my MBAs who run it into the ground as soon as there is enough of a client base to Sell. you fucking fucks all suck.
Similar to Memrise, which was really fun when you could make your own mems using imagesearch and customise everything… And now is a rubbish duolingo clone.