Duolingo CEO tries to walk back AI-first comments, fails
Duolingo CEO tries to walk back AI-first comments, fails
- Following backlash to statements that Duolingo will be AI-first, threatening jobs in the process, CEO Luis von Ahn has tried to walk back his statement.
- Unfortunately, the CEO doesn’t walk back any of the key points he originally outlined, choosing instead to try, and fail to placate the maddening crowd.
- Unfortunately the PR team may soon be replaced by AI as this latest statement has done anything but instil confidence in the firm’s users.
I mean, it is too late. Canceled my sub, won't be coming back.
Same. Deleted the app this weekend and let my 918 day streak evaporate.
I’m actually kind of surprised at how little it affected me, to be honest. I had a little bit pre-regret about losing the streak before deleting the app, but now a couple days later that feeling certainly doesn’t exist. AND there’s that benefit of no stupid owl guilt tripping you every day.
Check out "Language Drops" and "Rosetta Stone" if you're looking for replacements. They both have very different approaches to language learning (both from each other and from Duolingo), but their content is at the very least much better curated than Duolingo's.
I haven't gone out of my way to check but AFAIK neither of them is jumping on the AI-before-anything-else train.
I canceled my sub, but sadly not out of principle on the AI thing. I just accidentally hit the button that accepts an upgrade to the family plan and it didn't look like there was an easy way to undo it so I just killed the whole subscription.
Where is the discussion for replacing CEOs with AI? Seems like predicting market trends based off of historical data and managing corporate resources would be just the sort of thing that AI would be good at. Plus it would cost way less and not require massive bonuses nor ownership of the company.
Well, the crux of the problem is that AI is trying to approximate intelligence. That's not useful for a CEO.
boom
Respec the AI with lots of points into Charisma and Legal Dexterity. It also needs a large amount of Gold to get started.
“AI is creating uncertainty for all of us, and we can respond to this with fear or curiosity. I’ve always encouraged our team to embrace new technology (that’s why we originally built for mobile instead of desktop), and we are taking that same approach with AI. By understanding the capabilities and limitations of AI now, we can stay ahead of it and remain in control of our own product and our mission,” writes von Ahn.
Now please explain in more detail how this advice should be followed, practically, by someone you just fired because AI was cheaper. Give examples of how they can "stay ahead of it" so as to "remain in control of the product and mission" they are no longer employed to work on. How should they "embrace" this transition and "respond with curiosity" to no being newly unable to afford food or rent? "Uncertainty for all of us" my ass.
The former employees are now curious about how they will pay rent and eat, so there’s that.
In another thread someone told me you can buy gems or something to keep your streak going.
That would've made me uninstall long before his comments.
I'm a long time user of Duolingo and you earn plenty to give yourself the occasional streak freeze if you can't go two days without doing a lesson. It's not really as predatory as it sounds. It's nothing like pay to win type games.
Fuck Duolingo for the AI shit though, don't mistake me for a Duolingo simp thinking their blameless. It's just that the monetization is not as predatory as it sounds.
I have so many could probably keep a streak foing indefinitely without ever doing a lesson, but I'd need to log in every couple days to repurchase the streak freeze.
Ah, okay, thanks for the info! I've never used Duolingo so I genuinely don't know.
"Freezing" your streak is just silly, even if they offer it for free. Is this just for online clout, so you can brag (falsely) to others how long you haven't broken a streak?
If an alcoholic goes 10 years without drinking, then has a beer, the streak is broken. Doesn't mean you can't recover and improve, but it is what it is. It's dishonest to pretend it didn't happen, especially if you're comparing yourself to others...
I remember easily getting gems for free. Also the streak basically doesn't matter at all. What made me uninstall is the slow pace. It felt like I was stuck on the same words and topics forever. It felt like I was not actually learning anything, which if you've ever started learning a language if a formal setting, is very apparent.
It's worse than that.
Yes, you can pay for a streak freeze. If you don't, you'll probably find that you were given one for free anyhow. You'd have wasted your gems.
Yes, you can pay to undo a streak loss. It's more than paying for a freeze.
It'll give you multiple chances to pay for all that, too. If you're out for days and then come back, you can pay to fix your streak.
What is the point of a streak if you can just buy your way back to it?
Also, I had paid for the last couple years, which (IIRC) includes free streak freezes. It still asked if I want to pay for them. I'd say no, and find I had one anyhow, or a friend had miraculously given me one.
But during the last year (365 days) my streak was actually only at 190 or so because I'd used so many streak freezes that I got for free. I wasn't even trying to keep my streak.
When I finally let my streak die, the icon started trying to guilt trip me into coming back with horrible icons of Duo being sad, heartbroken, or even dead.
The constant mental manipulation that was well beyond what gamification should ever be was what finally drove me to just quit playing altogether. I had already canceled my sub long ago, but I'm not even going to use the remainder of this year I've already paid for.
Tl;Dr: skip the apps unless they're part of a bigger in-person course. Prefer reputable sources like pimsleur and mango languages. If you have no rush, get graded readers and watch a lot of YouTube, podcasts, etc.
Ok, so here are my two cents on learning languages and the whole category of learning apps. They are all flawed on some major way or another. But mostly it is about pacing learning progress.
Teaching absolute beginners is easy. They know nothing, thus anything you show them will be progress. The actual difficulty when learning a language is finding appropriate material for your level of understanding, such that you understand most of it, but still find new things to learn. This is known as comprehensible input. The difficulty of most apps is that they are not capable of detecting then adapting study content accordingly to the student's progress. So they typically go way too slow, or sometimes too fast. Leaving the student frustrated and halting learning.
Jumping with some nonzero knowledge into any app is also torture. It's known as the valley of despair. The beginner content is too boring and dull, now that you know a bit, but the intermediate level is way too much of a gap for you yet.
My advice is to skip language learning apps. The "motivation via gamification hypothesis" is flawed and lacks nuance and understanding of behavioral science. People don't stop studying out of a lack of tokens, gems, streaks or achievement badges. It's because the content itself is uninteresting and bores them. Sure, the celebration and streaks work at first, but they usually lose effect by something known as reinforcement depreciation. The same stimulus shown too much or too frequently stops being gratifying. The biggest reward for learning a language is actually using it.
A method that is known to work is to find graded readers. Watch a lot of YouTube, podcasts, social media, in the target language (avoid the language learning influencers) listen to native influencers speaking about topics you care about. Books work, in-person courses work, learning apps are good to start you up form absolute zero. But most learning happens on what you do in your everyday life. Using the language is the most effective way of becoming good at the language. Everything else is just excuses for using it.
exactly. I also don't appreciate the app changing the icon to guilt trip me back into their odd choice of/irrelevant vocabulary that I am supposed to learn
You put into words exactly how I’ve felt about language learning apps. Every time I try a game or app that's supposed to teach you, it feels like I’m starting over, and it never actually becomes fun. I tried Duolingo, but after a while, I found myself just doing super easy lessons to keep my streak going so I wouldn’t have to sit through the boring ones. It felt pretty bad, so I stopped using it when I hit 800 days.
My friend didn't use any apps and instead started by texting and talking with people and managed to learn Korean in just a year, well enough for casual, everyday conversations or hobby-related stuff. Meanwhile, I’ve been using apps and still can’t hold a conversation with anyone…
I deleted the app the second he said this. Get fucked, AI.
Classic "I've made a HUGE mistake" moment from yet another "thought leader" suffering from AI/layoff FOMO. 🙄
How do these people become CEOs they're as thick as several short planks nailed together.
Firstly every single company that has tried to replace its employees with AI has always ended up having issues. Secondly even if that wasn't the case, people are not going to be happy about it so it's not something you should brag publicly about.
If you're going to replace all of your employees with AI just do it quietly, that way if it fails it's not a public failure, and if it succeeds (it won't) then you talk about it.
How do these people become CEOs they’re as thick as several short planks nailed together.
Being a CEO has absolutely nothing to do with intelligence, I guarantee you that Duolingo has employees who are far more intelligent than the CEO.
If they do it quietly, they won't get the stock price bump every company gets from saying they're going to replace (costly) employees with AI.
People who are smart in one or two domains often overestimate how smart they are in other domains. They develop a mental model, confirm it quickly, and never re-asses it.
The issue with AI, is we're probably hitting our first real S curve with the current technology's performance but a lot of people who bet big are only see the exponential part and assuming there won't be a level off, or that the level of is far away.
There is no Moore's law for AI.
Intelligence has nothing to do with success. These people are born into wealth, are wealth-adjacent, or are expert ass-kissers.
They also tend to be more greedy than others for wealth, status and power, and not imaginative enough to see that life is about more than this. So they dedicate their life to crawling up to the top of the corporate heap while everyone else gets on with actual real stuff.
People keep forgetting that these companies’ product is stock price, not whatever they’re advertising at any given moment.
Their “CEOs” have gotten sloppy because the grift has gotten so easy they naturally assume everyone is in on it. If everyone is in on the grift, there’s no need to lie about it.
What is the point of this news that talk about a walk back that is doing nothing to walk back?
crazy how fast they ruined the reputation of this company. just a couple months ago, duo mascot and Duolingo streaks were cool and fun. they had a good thing going. but now it's just another shit tech company again. they lost all the good will in like a month.
crazy how fast they ruined the reputation of this company.
they lost all the good will in like a month.
Twitter enter the chat
Twitter was going downhill hard already when Musk bought the place
I uninstalled yesterday, good riddance Duo.
Last year in February I uninstalled the app on a perfect, 2000-day streak when I got the first whiff of AI; I'm probably never going back
Now say it in Spanish.
FTFY: Pretends to walk back his statements. Fools no one.
So...tries and fails? 😛
except his AI.
If anyone wants to practice their Japanese or have questions, they can message me.
主よ、如何なさいましたか。
I hear good things about Pimsleur as an alternative.
Interesting.
I have switched to Mango Languages because my library gives free access to it. So I’ve been trying to share that information with people. Or, at least, check your local library to see if they offer something.
That’s great advice. At least for as long as libraries still exist. 😫
Pimsleur has been making "real" language courses for 20+ years, you could get CD's with languages back then, there should be plenty to choose from.
It can go either way, some people like the method, others hâte it because it's not gamified. Pro tip, get pimsleur courses from your library if you want to try them for a real trial rather than what they give you
People are unfair with this "CEO". Its statement helped me move on from duolingo, which has seen significant decline in quality while never going beyond "a moderately bad way to start learning", toward better, more developed, more cared for, cheaper, solutions.
So, thanks for that.
Please do share :)
I'm mainly interested in Japanese, so I'm currently looking at https://www.renshuu.org/ . In addition to just throwing random stuff at you, it gots some more in-depth training, explanations of stuff (something that never happened in duolingo), additional hints for alphabets including some mnemonics, and years of dedicated experience in the language. I can't tell how it would feel long term, but so far even having some basic explanations is a great improvement.
I love this headline so, so much
I want every headline to end with "..., fails"
If you're gonna be a shitty company and do shitty things, just STFU. I don't know why they think they can talk their way out of it, it's like they believe their own BS.
What I'd wonder is why it's such massive expensive for Duolingo to hire 2 or 3 people to cover a language anyway. Presumably most of the work is contractual - hire somebody competent to produce a course, get somebody to say the lines, refine the course based on feed back and that's mostly it.
Terminal MBA brain
I would assume it is more about time than money. It is a big investment making a whole functional system with llm (I have a hard time believing it is actually AI), it will cost a lot if done wrong (just like everything else). You can't just prompt "make a course in Spanish" and get anything good out of it and you still need ppl who can quality check the output. I could see them use it to mass produce sentences and stories in different levels (not the actual story) and voice recordings, but not actually anything creative and I would assume that is the goal. But we have seen too many shitty products to believe in anything with llm.
boring, broken garbage content. monthly subscription fee. get the fuck out of here
I just started using Duolingo to learn Spanish. Can anyone recommend alternatives they have had success with that function the same way?
I tried out a bunch, including Babbel, Busuu, Language Transfer, Mango, and Memrise. I didn't like them for one reason or another. I finally landed on Lingodeer. It's similar to Duolingo, but it is a paid app. (You can try level 1 of any language for free.)
The regular subscription price is definitely not worth it. It's okay (not great, but not awful) when they do their sales. But I felt okay about paying human workers.
This kind of learning is a great start, but will only get you so far. If your local library has access to Kanopy, look for the Great Courses series on Spanish. I thought that was an excellent series after a little bit of Duolingo.
Anki is free. If you need gamification, then perhaps memrize is for you. I'd just go with anki though. Ankidroid is a good app to work with the anki decks.
Use free Anki and get a free 1k or 5k high-frequency community deck from Anki website. Or get Refold 1k deck (paid) for anki.
If you find Anki too complicated and you don't mind paying a sub (look for discount/vouchers), use lingvist (paid) or memrise (not sure how this app is now after the changes) to learn 1k words. Any app that focuses on high frequency vocab is fine I think.
Cancel subscription once you have learnt 1k words or can read comfortably a simple native book or graded books, or understand a podcast designed for learner (example InnerFrench), probably will take 1-3 months at about 10-30 words a day.
The main difference between 1k and 5k decks is that the 5k decks include very common type of words like "the", "a", "he", "she", "is", "are", which are so high frequency that you will acquire them by just doing anything in the language. Either type of deck is fine, it is up to you.
Try reading graded readers with audio at the same time as you are going through your deck so you are getting more context for new words you learn. You will encounter new words while reading before seeing them in the deck, which has a positive effect in remembering the word. Reading also helps serve to test how much you have improved in using the language.
Read up on some basic high frequency grammar in your target language. Depending on language you will have to also actively learn the alphabet, numbers, phonic and so on before doing any of the above.
The main idea of learning high frequency vocab is to start consuming content as soon as possible. Never forget that using(reading, listening, writing, speaking) the language is the main purpose of learning languages.
If you like gamification and keeping scores, count the books/article read, count the words learnt, count the hours spend listening don't count coins or gems.
Anki - https://apps.ankiweb.net/
AnkiDroid - https://play.google.com/store/apps/details?id=com.ichi2.anki
Anki shared decks - https://ankiweb.net/shared/decks?search=french
Refold decks - https://refold.la/category/decks/?show=all
Lingvist - https://lingvist.com/
There should be a federated system for blocking IP ranges that other server operators within a chain of trust have already identified as belonging to crawlers. A bit like fediseer.com, but possibly more decentralized.
(Here’s another advantage of Markov chain maze generators like Nepenthes: Even when crawlers recognize that they have been served garbage and they delete it, one still has obtained highly reliable evidence that the requesting IPs are crawlers.)
Also, whenever one is only partially confident in a classification of an IP range as a crawler, instead of blocking it outright one can serve proof-of-works tasks (à la Anubis) with a complexity proportional to that confidence. This could also be useful in order to keep crawlers somewhat in the dark about whether they’ve been put on a blacklist.
Any recommended alternatives?
Just say AI bad use is cancer for human kind. Good use can help humans to do their task with less effort. So its all depend on usage.
Duolingo is one of those fad apps that are overrated but are exquisitely trash once you force yourself to use them just to say "Hey, I have learned a language!". Honestly, it was pretty funny seeing Duolingo comment on videos on all that despite it feeling forced, replacing it with AI just signifies Duolingo's slow and painful demise