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

161–170 of 175 posts

Re: What Google Translate can tell us about vibecoding

#161

Post by anticensor »

mordechai9000 wrote: post #148
tanvach wrote: post #133

Earlier quoted context omitted.

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.

Swedish has informal as the default register instead.

Re: What Google Translate can tell us about vibecoding

#162

Post by insane_dreamer »

I've done freelance translation in two different language pairs, for 20 years (mostly as a side thing). A few thoughts on this:

- Google Translate has always been garbage for professional translation work, which is why human translators have been needed.

- LLMs can write in a way that sounds native (at least in English), which is something that ML translation software was bad at. They can also understand context to some degree and can adopt tones. This is a huge leap forward that makes it suitable for a large majority of translation work _between common language pairs_ (Hindi to Thai is probably safe).

- The vast majority of translation work is commercial in nature, and most cases, "decent" is good enough. Cost is prioritized over occasional mistakes. Prices have been falling for years already. Many agencies now pay peanuts because they expect you to use AI and "just" do some light proofreading. Except that often the "light proofreading" was actually "heavy editing" that often took more work than translating from scratch. It's not worth it.

- Do not lump translators and interpreters together when discussing this subject. Interpreters aren't going away and their job is much more difficult to replace with AI (for a variety of reasons). What they do is very different than translators (some do both of course).

- There are some exceptions to "decent is good enough", where: 1) context/localization is critical (a major advertising campaign), 2) accuracy of the message is critical (a government pronouncement), 3) linguistic excellence is critical (a novel). This may be the most visible parts, but also a very small fraction of the overall market.

My conclusion is that translators won't disappear altogether but their ranks will shrink considerably.

Re: What Google Translate can tell us about vibecoding

#163

Post by voxleone »

>>For what it’s worth, I don’t think it’s inconceivable that some future form of AI could handle context and ambiguity as well as humans do,

Absolutely! AI might get better at context, but there's still no substitute for a seasoned dev who’s had to untangle a 2 a.m. deployment gone wrong. As for vibe coders — hey, good vibes are great, but they won’t refactor a legacy monolith on their own.

Re: What Google Translate can tell us about vibecoding

#164

Post by rstuart4133 »

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…

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

Quote from article:

> it turns out the number of available job opportunities for translators and interpreters has actually been increasing.

Re: What Google Translate can tell us about vibecoding

#165

Post by tkgally »

djaychela wrote: post #143
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!

The combination, compilation, and evaluation process was designed by Claude Code and I haven’t examined or evaluated it closely yet. Here is the log of a session I did just now, after a few more hours of vibe-polishing the program:

https://www.gally.net/temp/20250619translationprogramtrace/i...

The Anthropic servers seem to have been overloaded, but if you read through the above it should give you an idea of how the program is currently doing that synthesis. If you want the code as it stands now (a single HTML file), feel free to email me. My website is linked from my profile page.

Re: What Google Translate can tell us about vibecoding

#166

Post by devnullbrain »

simonw wrote: post #21
devnullbrain wrote: post #16

>All this is not to say Google Translate is doing a bad job Google Translate is doing a bad job. The Chrome translate function regularly detects Traditional Chinese as Japanese. While many characters are shared, detecting the latter is trivial by comparing unicode code points - Chinese has no kana. The function used to detect this correctly, but it has regressed. Most irritatingly of all, it doesn't even let you corr…

That doesn't sound like a problem with Google Translate, it sounds like a problem with Google Chrome. I believe Chrome uses this small on-device model to detect the language before offering to translate it: https://github.com/google/cld3#readme

Archived in 2024

Re: What Google Translate can tell us about vibecoding

#167

Post by simonw »

devnullbrain wrote: post #166
simonw wrote: post #21

Earlier quoted context omitted.

That doesn't sound like a problem with Google Translate, it sounds like a problem with Google Chrome. I believe Chrome uses this small on-device model to detect the language before offering to translate it: https://github.com/google/cld3#readme

Archived in 2024

Looks like it's still vendored by Chromium: https://github.com/chromium/chromium/tree/main/third_party/c...

Re: What Google Translate can tell us about vibecoding

#168

Post by coolpool »

I've just launched https://www.verbascan.com/ — it does context-aware photo translations. Still early (doesn’t ask questions yet), but even at this stage, it's clear how much better the results are when intent and nuance are factored in. Would love any feedback if you give it a try.

Re: What Google Translate can tell us about vibecoding

#169

Post by Alex-Programs »

tkgally wrote: post #112
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 nice! Thanks for the links.

I just did some more research, which you might find interesting. Thinking actually makes them translate worse!

https://nuenki.app/blog/the_more_llms_think_the_worse_they_t...

Re: What Google Translate can tell us about vibecoding

#170

Post by tkgally »

Alex-Programs wrote: post #169
tkgally wrote: post #112

Earlier quoted context omitted.

Very nice! Thanks for the links.

I just did some more research, which you might find interesting. Thinking actually makes them translate worse! https://nuenki.app/blog/the_more_llms_think_the_worse_they_t...

Also very interesting! Excellent research design and presentation, too.

Your results accord with my own (much less systematic) tests of the translation of short texts by reasoning models. The issue becomes more fuzzy with the translation of longer texts, where quality is more difficult to evaluate objectively. I'll drop you an email with some thoughts.

Post reply on HN