1,298 karma · joined May 18, 2022
Each one of the roles you listed above is only passable with AI at a superficial glance. For example, anyone who actually reads literature other than self-help and pop culture books from airport kiosks knows that AI is terrible at longer prose. The output is inconsistent because current AI does not understand context, at all. And this is not getting into the service costs, the environmental costs, and the outright intellectual theft in order to make things like illustrations even passable.
Besides, I'd rather have some more word pronunications than memorizing a table of der/das/die,dem/dem/der and a word's gender on top of learning the word itself. Or changing the position of a verb depending on if I used a modal verb or not.
It is not "beyond fucked" that things have different pronunciations sometimes. Other languages have problems for people who solely learn by speaking. It's not unique to english.
A great deal of things are already packaged, but for the most part I find it pretty fast to package something. Once you write a derivation [2] or two, it's not that bad. I never packaged for other distros because they all seemed quite tedious, but the nixpkgs reference [3] lists most things and I can look at the source of similar packages in nixpkgs. It is a time commitment though to learn so it's understandable that it's not really appealing.
[0]: https://github.com/svanderburg/nix-patchtools
[1]: https://blog.thalheim.io/2022/12/31/nix-ld-a-clean-solution-...
"i don't use Nix partly because all my friends who use Nix have even weirder bugs than they already had and partly because i don't like the philosophy of not being able to install things at runtime."
The first part is mostly true. Nix installs things in a readonly store (/nix/store) so regular dynamically linked binaries don't work. Packaging takes a different approach and when things break, it can be difficult to work around. That said, I've run NixOS for over a year now and I find the benefits are far preferable to these downsides. It's not often I run into bugs, let alone show-stopping ones. What is annoying is how many tools are distributed without the source, so I have to run patchelf on them or use something like nix-ld.
As for the latter part, I think that using Nix will change that mentality. (Note that you can do `nix-env -iA $pkg` but it's not recommended). See, I don't even install things like rust at a global level anymore. I can always do `nix-shell -p $pkg` for an ephemeral shell if I need that, or I encode that dependency directly in the project's flake.nix. If I end up using that program a lot I will make the effort to add it to my NixOS config.
Took less than a minute to write, test, and get the right dependencies (nix-shell -p imagemagick ghostscript). Scales to any number of pdfs.
The terminal just wins probably 95% of the time as long as you know about available tools.
The song is adequate. The beat, drop, and chords are not that complicated, you could've learned how to do that yourself and not be chained to suno or whatever.
I would be delighted if a friend took the time to personally write and produce a song about me. I would not care at all if I realized it was just auto-generated stuff being pumped out "5x per week". The former I would cherish, the latter would be disposable junk to be listened to once and then forgotten about.
It's slop because it doesn't mean anything or has any value, not because it's literally sloppy.
---
"listening to their same 20-40 artists who would be in billboards list each month"
That's a you problem, quite literally a skill issue. There are so many indie artists making wonderful music that's leagues better than whatever hyper-distilled swill is on the radio at the moment.
---
"tell me it has hurt the abundance and diversity of songs out there"
Actual artists who make music composition and production their career ARE hurt when they can't find jobs/gigs/commissions because of things like suno. These tools are predicated on wholesale theft of music produced by humans.
But you can generate disposable songs at breakneck pace that don't mean anything, so it doesn't matter.
Zero-effort output generators like prompting means people are just generating trash that they themselves don't even care about. So why should I take my time to watch/experience that?
The whole "GenAI is accessible" sentiment is ridiculous in my opinion. Absolutely nothing is stopping people from learning various art mediums, and those are skills they'll always have unlike image generators which can change subscription plans or outright the underlying model.
Absolutely no one should be lauding being chained to a big corp's tool/model to produce output.
---
Why should being an artist be a viable job? Well, people should get paid for their work. That applies to all domains except technical people love to look down on art while still wanting to watch movies, well produced youtube videos, etc. You can see it in action here on HN frequently: someone will link a blog post they took time to write and edit... and then generate an image instead of paying an artist. They want whatever effect a big header image provides, but are not willing to pay a human to do it or do it themselves. Somehow because it's "just art" it's okay to steal.
---
If tech has progressed to the point of true "general artificial intelligence", then likely all jobs will be obsolete and we're going to have to rethink this whole capitalism thing.
I think all industries should be utilizing tech to augment their capabilities, but I don't think actual people should be replaced with our current guesstimator/bullshitter "AI". Especially not critical roles like doctors and nurses.
While I agree that everyone has different opinions, I do think the average person will appreciate being able to look back after so many years instead of having nothing due to repeating some mantra without thinking about it.
The thing is that it requires a lot of time and studying. At least 30 mins per day is the bare minimum, but 1-3 is much better for results.
I don't personally feel like language learning is easier for me than other people. I just focused on putting the time and studying in diligently over many years.
The problem is that it's grossly inefficient time-wise, and the content of "conversations" was always very, very simple. "Hi my name is _, I like the color _, My hometown is in _, how are you today?" Is not a real conversation. It's boring and most students learn the vocab for the upcoming chapter's test, then forget it after.
I'll concede that with 3 semesters of German, were I to pick it up again, I would probably do so pretty quickly given that the teachers paid a lot of attention to our essays.
It's probable that small classes would help because the teacher could then be more of a private tutor. But with 20-30 size classes, only really motivated students who already study/watch media outside of school will excel. So it's kind of redundant in my opinion.
Diligent self-study with attending a language exchange or another environment to speak/practice the language will yield much greater results much faster. You can study the same textbooks at your own pace, you can find additional material and study groups, and you can hire a tutor at times to fill in gaps.
I think if you're a college student it's fine since you have to pick a class anyway (I had to take 3 semesters of any language), but as an adult where time is significantly more precious, I can't recommend it. In a sibling comment I went over what I do use.
1. SRS - Spaced Repetition Software, for flashcards. Anki is the gold standard. It's open source and free on every pc/android/etc except iphone where it's $20 I think. I recommend finding a good starting deck with about 3k to 6k words to help build your core vocabulary. In my case it was "Evita's 5k Korean". For about 6-8 months I grinded 20 new words per day, which means about 30-50 minutes of Anki depending on if you missed a day or not and thus had a backlog. If you have less time I recommend 5 or 10 new words per day.
2. Find trusted resources for grammar and structured learning. You might have to hunt around but for Korean, I found some excellent websites, Youtubers, and textbooks like Korean Grammar in Use I-III. These materials really are the core of your studying. Vocab doesn't help much if you don't know grammar and you certainly can't say anything without vocab. These are how you get to output, i.e. writing and speaking correctly.
3. Find graded readers if possible. Roughly, these are texts designed around 90% comprehension which is a sweetspot for learning new words naturally through context. Unfortunately at the time I couldnt find any for Korean, but I've watched friends use them for e.g. Mandarin Chinese and learn quite a lot of vocabulary in a short time.
4. Find someone who can correct your writing in some form. Whether that's a private tutor or a friend who's native language is your target language and their target language is your native language. In the past I found some dedicated learners through HelloTalk who would trade journal entries with me. I would correct their English and they would correct my Korean. It goes without saying that you need to practice output in your target language when possible, both in writing and in speech.
5. Find a good language exchange and/or friends who speak your target language. By good, I mean a structured language exchange that enforces pairings and language usage. In Seoul I find that most "language exchanges" are excuses to drink and and chat, mostly in English. There was one language exchange that 1:1 Korean language-only pairings for 1 hour, then I repaid that with 2-3 30minute pairings of 2-3 people in English. This is where you put your textbook/solo studies to practice by actually speaking (and hopefully getting corrections). Eventually I hit a plateau and got tired of having similar conversations, plus paying $10 per event. I also found a few lifelong friends who are studying English and thus we can ping each other for random questions.
6. Find some spaces or groups that are -only- in your target language. With the internet it's easier than ever now with Discord. For example, my friend learned a lot of French by hanging out in French speaking gaming servers on discord. There are also apps like Hilokal and HelloTalk, but I haven't used them in a while so I can't speak to their quality anymore. Lastly there are offline options depending on your area. In the US I used Meetup to find language groups and in Korea I use, well, a korean equivalent to find groups in niches I enjoy.
7. Lastly, and this isn't a tool, but "If you want to improve, be content to be thought foolish and stupid." - Epictetus. In learning a language, you will make mistakes and you will say things that sound stupid. It's okay. It's unavoidable and you make good progress by learning from these mistakes, so long as you reflect on them and understand why the mistake occurred. The people who focus on being perfect and making zero mistakes in learning a language, in my experience, do not go very far.
My background is that I've studied Korean for ~8 years now, as a native English speaker. Like most US citizens I took Spanish classes in middle & high school. I did the traditional classroom method with 3 semesters of German in college. And I forgot most of Spanish and German aside from some words and grammatical rules, because neither got me to a level of conversations with native speakers or being able to engage with media.
Duolingo and most classrooms (I know there are exceptional curriculums and exceptional students) don't prepare you to actually speak to people. They prepare you to engage within their systems, aka answering tests or whatever. This is not speaking a language but moreso learning about it academically.
There is a lot to discuss but I've never been able to recommend Duolingo, even before they reduced their staff and replaced them with AI. Why? Because it's inefficient with regards to your time, and the content is too insubstantial. It's possible to spend a year of your time on Duolingo and barely be able to speak the language at all with someone... which is kinda the whole point of studying a language?
I love the hobby of studying languages and things like Duolingo and the classroom method put people off when they can't speak very much even after a long time investment, which is damn shame.
My point is neither should really be looked towards for substantial language learning methods.
"Most common chord" is mildly interesting, but not really that useful. The most common key, and the most commonly used chords relative to that key (i.e. with roman numeral analysis) would be much more useful and interesting. This would help paint a clearer distinction between e.g. country and jazz, not that "jazz uses Bb major more". Also, anyone with general instrument knowledge would surmise that since Bb and Eb instruments are much more prevalent.
"If you’re sitting down to write a song, throw a 7th chord in. The ghost of a jazz great will smile on you."
7ths don't belong to jazz only, and the average songwriter isn't making data-driven decisions on how to settle on the chord structure for their song.
I'd MUCH rather a hobby web server be written in rust than C++ or C. I'm not touching the latter with a ten foot pole.
No one thinks this except some strawman that you've devised. No point in reading anything else in this comment when this is so blatantly absurd and detached from reality.
* Obviously the works need to be readable in a language one knows. But it's not like the essence of literature classes change whether one speaks English or German or whatever. That's not the point.