rzz3

It doesn’t have to be that way; learning a language is a long process. I took about 3 years to reach real practical fluency, and I still have to say “what?” more often than I’d like and I still need to learn a ton of more advanced vocabulary. Duolingo unfortunately doesn’t offer any C1 content so I’m stuck using other methods.

I disagree. In Spanish, learning the subjunctive is essential and that’s part of B2, and I think Duolingo did a good job of teaching it. If you can’t understand “Que te vaya bien” even completing a purchase at a store would be a bit difficult.

So I agree they go over the top with it, _but_ I reached fluency in Spanish in about 2.5 years and Duolingo was an indispensable part of it.

> if the users aren't actually becoming proficient by using the app

Learning a language to fluency requires real commitment, and I’d say an app could never possibly do it on its own. One of the most key things Duolingo gave me was consistency and a lack of an excuse to constantly practice and learn. But you also have to (and I did) use the language daily, watch content in the target language, travel and speak with locals in the language, etc. I’m not sure where Duolingo ever claimed that it alone was enough to actually reach proficiency or fluency.

Duolingo’s gamification and streaks and leaderboards gave me a reason to put a lot of effort into learning the language, and I don’t know where I’d be without it. There’s a lot of things about Duolingo I don’t love but I’m incredibly grateful that it exists.

What do you think the correct, or at least a more likely tree is?

Another company I’ll never want to work for then.

I don’t understand why we’re not going to nuclear.

I also jumped out of the funnel and had this “I got tricked” feeling after selecting the story parameters. I’m someone who would use this, too. But I’m not giving you my personal information unless I can see what it does.

I don’t want to be negative, but writing a mathematical proof and planting a tree are about the worst metaphors I could possibly think of. Building a house is probably a better example for something that could be infinitely maintained and improved.

After the bad metaphors, all the article really says is “do visible work and finish it”. This seems obvious.

Here’s one from January: https://momen.app/blogs/what-is-vibe-coding-pros-cons/

Lol, sigh. Author says the term was coined 84 days ago on February 6th 2025. Literally go to Google and search ‘“vibe coding” before:2025-02:01; I see posts from more than a year ago.

I agree with you apart from WhatsApp. I automatically know you’re from America if you don’t use WhatsApp.

I’ve always felt like (as neither a mathematician nor a physicist) that “dark matter” is simply just something that suddenly makes a math problem work to model the universe-—and that in reality, that math problem just doesn’t work.

Is my theory even _possible_ here, or am I missing something. Really fundamental?

Thanks for the explanation, it makes a lot of sense. One thing I keep wondering though is why we can’t teach them logic at some sort of base layer. Not to anthropomorphize, but couldn’t the actual next-token prediction itself embed some sort of consultation against another (non-LLM) model that can evaluate a probability of logicality of the overall output that would be produced be selecting next-token-A vs next-token-B? I’m sure it’s obviously way more complicated than this, but still.

Does anyone have any technical insight on what actually causes the hallucinations? I know it’s an ongoing area of research, but do we have a lead?

Mainly because I see moral alignment and I see Sam Altman as a person of good moral standing. I don’t see any perception of morality from Google, just a faceless mega corporation.

I acknowledge that it’s more of a perception issue than anything, but my _perception_ is that I don’t trust Google as far as I can throw it.

How would you say they’re solutions in search of a problem? I engage LLMs more than 50 times a week to solve problems at work, in my personal projects, and in my personal life. I respect your opinion but thoroughly disagree.

Just for my own context and perspective, do you use generative AI or not?

Our current advances in LLMs might just represent the biggest achievement humans have made since discovering fire.

> OR to add regulation around infinite video scrolling.

I really don’t want the government telling me what I can or can’t do on my phone, or that an app I enjoy can’t exist. Alcohol exists, gambling exists, cigarettes exist, porn exists, cars can drive fast, and yet because I have self control and good judgement, I haven’t allowed any of those things to get a hold over me either. I don’t want the government to be my dad. And even if you did, can you really trust our technophobic corrupt out-of-touch lawmakers to get such regulation right? These are consumption-side problems in my opinion, and individuals need to bear the responsibility rather than trying to pawn it off on big tech companies or regulators.

I saw Claude 3.7 write a comment in my code in Russian followed by, likely from a previous modification, the English text “Russian coding” for no reason.

You really hit the nail on the head with trust. Knowing the power of these AIs and how absolutely little I trust Google, I’d never tell trust Gemini with the things I’ll say to ChatGPT.

I’d say priorities and values are pretty different, but there can be some overlap. But the problem is, folks don’t give it the required amount of nuance, and simply loop in all the horrible things the other party does and stands for with the values of that person, and it’s usually not accurate.

> In this world, asking "who did you vote for" isn't a matter of tribalism, but it is a (good) proxy for asking someone "what are your values".

I strongly disagree. In this duopoly of a political system, most people on both sides are just picking the lesser of two evils. Meanwhile, we are creating an alarmingly decisive political society by choosing not to associate with those who vote differently than us. Perhaps most importantly, we lose the opportunity to actually shift the political positions of others (and ourselves) by not engaging in healthy and non-judgmental political discussions with our friends and neighbors, ultimately increasing polarization even further.

Not everyone is voting based on their values—some are simply voting their wallets or the special interests they align with. Someone who is pro-choice, pro-LGBT, and pro-immigration may very well vote Republican because they work in the US Automotive industry, and so do their friends and families and people who they care most about. It doesn’t necessarily mean their core values are different than yours, but instead maybe simply just their priorities.

Passkeys do definitely make the mobile web experience better, but unfortunately they’re still not widely supported. I’m not saying mobile web apps can’t be good, but a native app allows for a lot of UX optimization.

I almost always book via apps. I can compare flights by looking at Kayak (app), then actually book it in the carrier app. I think the workflow just has to adapt to the tools you’re using, and trying to follow the same methods you’d use on desktop just don’t work. I don’t think either particular method is objectively worse than the other for every use case.

460km/h seems close enough to count for the article’s prediction, especially if the mentioned population difference is considered acceptable. IMO predicting the future shouldn’t just be about predicting the future of the USA. Many countries are more advanced in their respective ways than the US.

Im sorry. I really just can’t understand or relate to this at all. Mobile web still feels like such a terrible experience, and apps generally don’t. When’s the last time you tried booking a flight on mobile web? And how do you deal with all of the real estate the browser steals? Having to log in every time when the app can just cache my authentication and FaceID me?

It’s simply not true. Not in any place I’ve ever seen in all the years I’ve been in Mexico.

> OK does not convey the same thing. Fine. Is "10-4" or "copy that" or "Roger" acceptable? Because the whole point of that text, in a work context (and this is always a work conversation; I like my partners but I'm not personal friends with them), is to convey "I have received your message".

So I had missed the “in a work context”, where I think your point is slightly more valid.

>And a reaction generates an SMS that is, until you realize what it is, kind of incomprehensible. "Loved '[my comment]'" makes no sense when you get it as a long comment.

Only if you don’t both have iPhones, and the necessary information is all at the beginning. You see “Loved” and a few words from the start of the message and you remember what you said and you know which message.

> The reactions may not help you I have trouble seeing how they help anyone. You made an obvious joke. A guy who laughs at anything reacts "HA-HA!!" or whatever. What is the contribution?

So this answer may surprise you, but emoji reacts actually do communicate something unique, and of value. They are more subtle. They also permit the conversation to terminate after the react. In my mind, and the mind of many others, an actual message necessitates a response, and I’d see it as rude to neither respond nor react to a message. Reactions are subtle and out of hand, and you don’t respond _to_ a reaction. Using a react instead of an actual text response has a way of signaling the other user that they don’t need to respond further, that you read their message, how you feel about it, and no further information is necessary. This is incredibly useful and I’d struggle to communicate with any kind of emotional nuance in this day and age without it. More fundamentally, reactions solve the problem of non-verbal communication and body language in text.

>This isn't "get off my lawn", much as I do love that.

It gives “get off my lawn” a bit when you say you don’t see how it would be useful to anyone. I’d never try to convince you that you need to use it, but I do think you should understand the utility for everyone else.

But, I definitely understand that it’s not useful for your professional use case as an anesthesiologist. Also, wow, what a cool job (seriously)!