What Google Translate can tell us about vibecoding
ingrids.space
What Google Translate can tell us about vibecoding
1–10 of 175 posts
Re: What Google Translate can tell us about vibecoding
#2In my limited experience, LLMs can have issues with translation tone — but these issues are pretty easily fixed with good prompting.
I want to believe there will be even more translators in the future. I really want to believe it.
Re: What Google Translate can tell us about vibecoding
#3Fascinating thought piece. While I agree with the thrust of the piece: 'that llms can't really replace engineers', unfortunately the way the industry works is that the excuse of AI, however grounded in reality has been repurposed as a cudgel against actual software industry workers. Sure eventually everyone might figure out that AI can't really write code by itself - and software quality will degrade..
But unfortunately we've long been on the path of enshitification and I fear the trend will only continue.
If google's war against its own engineers has resulted in shittier software - and things start break twice a year instead of once - would anyone really blink twice?
Re: What Google Translate can tell us about vibecoding
#4This seems like a terrible comparison since Google Translate is completely beat by DeepL, let alone LLMs. (Google Translate almost surely doesn't use an LLM, or at least not a _large_ one given its speed)
Re: What Google Translate can tell us about vibecoding
#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 good. I say this as someone who worked for twenty years as a freelance translator of Japanese and who still does translation part-time.
Just yesterday, as it happens, I spent the day with Claude Code vibe-coding a multi-LLM system for translating between Japanese and English. You give it a text to be translated, and it asks you questions that it generates on the fly about the purpose of the translation and how you want it translated--literal or free, adapted to the target-language culture or not, with or without footnotes, etc. It then writes a prompt based on your answers, sends the text to models from OpenAI, Anthropic, and Google, creates a combined draft from the three translations, and then sends that draft back to the three models for several rounds of revision, checking, and polishing. I had time to run only a few tests on real texts before going to bed, but the results were really good--better than any model alone when I've tested them, much better than Google Translate, and as good as top-level professional human translation.
The situation is different with interpreting, especially in person. If that were how I made my living, I wouldn't be too worried yet. But for straight translation work where the translator's personality and individual identity aren't emphasized, it's becoming increasingly hard for humans to compete.
Re: What Google Translate can tell us about vibecoding
#6The distinction between what people typically imagine a translator's job is and the reality reminds me of pixar movies being "localized" instead of just translated (green beans on a plate in the Japan release instead of broccoli because that's the food that Japanese kids don't like).
Lacking cultural context while reading translated texts is what made studying history finally interesting to me.
Re: What Google Translate can tell us about vibecoding
#7This seems like a terrible comparison since Google Translate is completely beat by DeepL, let alone LLMs. (Google Translate almost surely doesn't use an LLM, or at least not a _large_ one given its speed)
For Google's Cloud Translation API you can choose between the standard Neural Machine Translation (NMT) model or the "Translation LLM (Google's newest highest quality LLM-style translation model)".
https://cloud.google.com/translate/docs/advanced/translating...
DeepL also has a translation LLM, which they claim is 1.4-1.7x better than their classic model: https://www.deepl.com/en/blog/next-gen-language-model
Re: What Google Translate can tell us about vibecoding
#8> ... 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…
Try Soniox for real-time translation (interpreting). With the limited context it has in real-time, it's actually really good.
Disclaimer: I work for Soniox.
Re: What Google Translate can tell us about vibecoding
#9Fascinating thought piece. While I agree with the thrust of the piece: 'that llms can't really replace engineers', unfortunately the way the industry works is that the excuse of AI, however grounded in reality has been repurposed as a cudgel against actual software industry workers. Sure eventually everyone might figure out that AI can't really write code by itself - and software quality will degrade.. But unfortunat…
Maybe AI can't replace engineers but it surely can apply downward pressure on engineers' salaries.
Re: What Google Translate can tell us about vibecoding
#10> ... 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…
Try Soniox for real-time translation (interpreting). With the limited context it has in real-time, it's actually really good. https://soniox.com Disclaimer: I work for Soniox.
I’ve been looking for that! Thanks