Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

What Google Translate can tell us about vibecoding

ingrids.space

Post reply on HN
Print view View original

141–150 of 175 posts

Re: What Google Translate can tell us about vibecoding

#141

Post by wil421 »

dlisboa wrote: post #140
falcor84 wrote: post #115

Earlier quoted context omitted.

Is that beans example a real thing? If so, I would hate for my kids to be subjected to that. The best thing about watching films from another country is that you're exposed to the culture of that foreign place and learn about how it's different from yours - I don't see why we'd try to localize away the human experience as if these differences don't exist.

Localization is more important than you might think. My wife worked for a company that helped provide teaching content for schools throughout Brazil. They'd interview teachers all over the country and one of the complaints from teachers in isolated communities was that they had to use the same textbooks as other places in Brazil without any regard to their own situation. They reported that many examples for starting…

They don’t have green beans in the Midwest?

We bought kids toys on Amazon and the fruits were strange. Not sure if they were Asian varieties or just made up.

Re: What Google Translate can tell us about vibecoding

#142

Post by crazygringo »

f38zf5vdt wrote: post #122
tkgally wrote: post #5

> ... a translators’ and interpreters’ work is mostly about ensuring context, navigating ambiguity, and handling cultural sensitivity. This is what Google Translate cannot currently do. Google Translate can't, but LLMs given enough context can. I've been testing and experimenting with LLMs extensively for translation between Japanese and English for more than two years, and, when properly prompted, they are really go…

It's funny -- I independently implemented the same thing (without vibe coding) and found it doesn't actually work. When I ended up with was a game of telephone where errors were often introduced and propagated between the models. The only thing that actually worked was knowing the target language and sitting down with multiple LLMs, going through the translation one sentence at a time with a translation memory tool w…

That doesn't match my experience at all. Maybe it's something to do with what your prompts are asking for or the way you're passing translations? Or the size of chunks being translated?

I have been astounded at the sophistication of LLM translation, and haven't encountered a single false-friend example ever. Maybe it depends a lot on which models you're using? Or it thinks you're trying to have a conversation that code-switches mid-sentence, which is a thing LLM's can do if you want?

Re: What Google Translate can tell us about vibecoding

#143

Post by djaychela »

tkgally wrote: post #5

> ... a translators’ and interpreters’ work is mostly about ensuring context, navigating ambiguity, and handling cultural sensitivity. This is what Google Translate cannot currently do. Google Translate can't, but LLMs given enough context can. I've been testing and experimenting with LLMs extensively for translation between Japanese and English for more than two years, and, when properly prompted, they are really go…

>creates a combined draft from the three translations

How is this part done? How are they chosen/combined to give the best results? Any info would be appreciated as I've seen this sort of thing mentioned before, but details have been scant!

Re: What Google Translate can tell us about vibecoding

#144

Post by throwaway2037 »

yoden wrote: post #124

Machine translation is a great example. It's also where I expect AI coding assistants to land. A useful tool, but not some magical thing that is going to completely replace actual professionals. We're at least one more drastic change away from that, and there's no guarantee anyone will find it any time soon. So there's not much sense in worrying about it. A very similar story has been happening in radiology for the p…

    > Machine translation is a great example. It's also where I expect AI coding assistants to land. A useful tool, but not some magical thing that is going to completely replace actual professionals.
I can say from experience that machine translation is light years ahead of 15 years ago. When I started studying Japanese 15 years ago, Google Translate (and any other free translator) was absolutely awful. It was so bad, that it struggled to translate basic sentences into reasonable native-level Japanese. Fast forward to today, it is stunning how good is Google Translate. From time to time, I even learn about newspeak (slang) from Google Translate. If I am writing a letter, I regularly use it to "fine-tune" my Japanese. To be clear: My Japanese is far from native-level, but I can put full, complex sentences into Google Translate (I recommend to view "both directions"), and I get a reasonable, native-sounding translation. I have tested the outputs with multiple native speakers and they agree: "It is imperfect, but excellent; the meaning is clear."

In the last few years, using only my primitive knowledge of Japanese (and Chinese -- which helps a lot with reading/writing), I have been able to fill out complex legal and tax documents using my knowledge of Japanese and the help of Google Translate. When I walk into a gov't office as the only non-Asian person, I still get a double take, but then they review my slightly-less-than-perfect submission, and proceed without issue. (Hat tip to all of the Japanese civil servants who have diligently served me over the years.)

Hot take: Except for contracts and other legal documents, "actual professionals" (translators) is a dead career at this point.

Re: What Google Translate can tell us about vibecoding

#145

Post by alganet »

tptacek wrote: post #36
alganet wrote: post #32

Earlier quoted context omitted.

> everything that is today a spreadsheet was a program some programmer had to write That is incorrect, sir. First, because many problems were designed to fit into spreadsheets (human systems designed around a convenient tool). It is much more likely that several spreadsheets were _paper_ before, not custom written programs. For a lot of cases, that paper work was adapted directly to spreadsheets, no one did a custom…

I'm not sure we're really disagreeing about anything here. If you think spreadsheets didn't displace any programmers at all, that's contrary to my intuition, but not necessarily wrong --- especially because of the explosion of extrinsic demand for computer programming.

You say spreadsheet software replace programmer.

I say spreadsheet software replace paper.

That's the disagreement. You have intuition, I have sources:

https://en.wikipedia.org/wiki/Spreadsheet#Paper_spreadsheets

https://en.wikipedia.org/wiki/Spreadsheet#Electronic_spreads...

Re: What Google Translate can tell us about vibecoding

#146

Post by Alex-Programs »

throwaway2037 wrote: post #144
yoden wrote: post #124

Machine translation is a great example. It's also where I expect AI coding assistants to land. A useful tool, but not some magical thing that is going to completely replace actual professionals. We're at least one more drastic change away from that, and there's no guarantee anyone will find it any time soon. So there's not much sense in worrying about it. A very similar story has been happening in radiology for the p…

> Machine translation is a great example. It's also where I expect AI coding assistants to land. A useful tool, but not some magical thing that is going to completely replace actual professionals. I can say from experience that machine translation is light years ahead of 15 years ago. When I started studying Japanese 15 years ago, Google Translate (and any other free translator) was absolutely awful. It was so bad, t…

Also, Google Translate is really not a particularly good translator. It has the most public knowledge, but as far as translators go it's pretty poor.

DeepL is a step up, and modern LLMs are even better. There's some data here[0], if you're curious - DeepL is beaten by 24B models, and dramatically beaten by Sonnet / Opus / https://nuenki.app/translator .

[0] https://nuenki.app/blog/claude_4_is_good_at_translation_but_... - my own blog

Re: What Google Translate can tell us about vibecoding

#147

Post by the_af »

abrugsch wrote: post #138
the_af wrote: post #118

Earlier quoted context omitted.

Same. I don't understand why people want (them and their kids) to be isolated from other cultures. If you're watching a movie set in Japan, India or China, have it be about their culture. If there's something you or your kids don't understand, make an effort to learn about it (and the green peas thing seems trivial to understand). Netflix also does something absurd with their subtitles. I was watching "The Empress" (…

Just saying "Pixar movies" was probably not a great example. They can be deliberately location ambiguous (Monsters Inc., Toy Story - though it's clearly _somewhere_ in America, The Incredibles - a generic "metropolis"/50's futuristic city, lightyear, Elemental) or very specifically somewhere (cars - mashup of Route 66 towns, Finding Nemo - Sydney when on land, Ratatouille - Paris, etc...) It makes sense to "translate…

Good point.

The localization of The Incredibles in Argentina was embarrassing, though someone must have thought it was a good idea. They used local voice actors popular at the time (though not so much today) with strong porteño (Buenos Aires') accents. They also referred to the streets with Argentinian names, e.g. "let's turn that corner of Corrientes Avenue!". The problem is that Corrientes Av is very typical of Buenos Aires, but nothing on screen looked anywhere close to it, so the whole thing was ridiculous and embarrassing, sort of like if the characters pretended they were in Tokyo and were Japanese.

What if they had gone the extra mile (maybe possible in the near future) and reskinned every character to look more Argentinian, and rethemed the city to look more like Buenos Aires, would I have been happier? Certainly not -- I want to see the original movie as intended, not some sanitized version designed to make me feel "at home" and not challenge me in the slightest.

(I watched the movie in English as well, mind you).

Re: What Google Translate can tell us about vibecoding

#148

Post by mordechai9000 »

tanvach wrote: post #133
Alex-Programs wrote: post #102

Earlier quoted context omitted.

I've ended up doing a lot of research into LLM translation, because my language learning tool ( https://nuenki.app ) uses it a lot. I built something kinda similar, and made it open source. It picks the top x models based on my research, translates with them, then has a final judge model critique, compare, and synthesise a combined best translation. You can try it at https://nuenki.app/translator if you're interested…

Very neat, love how there’s a formality level selection! Google translate has such bad tendencies to use very formal language (at least when translating into Thai) that it’s almost useless in real life. Some English to Thai examples I tried so far have been quite natural.

I assumed Google errs on the side of formality because being informal in an inappropriate context is worse than being too formal for someone who is obviously not a native speaker. Not for Thai in particular, just in general.

Re: What Google Translate can tell us about vibecoding

#149

Post by jvanderbot »

roxolotl wrote: post #62
jvanderbot wrote: post #44

Earlier quoted context omitted.

When ordering, "I will have" sounds reasonable. When asking someone to pass them to you, just imagine them turning to you, looking you in the eye, and asserting "I will have the potatoes" like it's some kind of ultimatum. Yes, that is strange.

It’s such an anachronistic statement I laughed out loud reading your comment. I even was taught that you can’t pass things mid air. You place the potatoes down between each person required to pass them to the person who wants them.

A non sequitur perhaps?

Re: What Google Translate can tell us about vibecoding

#150

Post by f38zf5vdt »

crazygringo wrote: post #142
f38zf5vdt wrote: post #122

Earlier quoted context omitted.

It's funny -- I independently implemented the same thing (without vibe coding) and found it doesn't actually work. When I ended up with was a game of telephone where errors were often introduced and propagated between the models. The only thing that actually worked was knowing the target language and sitting down with multiple LLMs, going through the translation one sentence at a time with a translation memory tool w…

That doesn't match my experience at all. Maybe it's something to do with what your prompts are asking for or the way you're passing translations? Or the size of chunks being translated? I have been astounded at the sophistication of LLM translation, and haven't encountered a single false-friend example ever. Maybe it depends a lot on which models you're using? Or it thinks you're trying to have a conversation that co…

I'm using o3 and Gemini Pro 2.5, paying for the high tier subscriptions. The complaints I get are from native speakers -- editors and end consumers. The LLMs tend to overfit to the English language, sometimes make up idioms that don't exist, use false friend words (especially verbs), directly translate English idioms, and so on. I've translated several book length texts now and I've seen it all.
Post reply on HN