There was doom and gloom about Mozilla saying they wanted to integrate ‘AI’ into Firexox, but to me, privacy respecting locally run translation seems like a great usecase. I guess people were only reading the headline and jumped to the conclusion Mozilla wanted a chatbot.
Features like this, better text to speech, audio descriptions for images, maybe something to find alternate sources covering the same news, spotting fake reviews, etc all seem like worthwhile developments.
They have been working on the local translation stuff since 2019, long before they started talking about AI recently
True, but a lot of stuff that’s now called ‘AI’ wouldn’t have been branded as such a few years ago. A lot of the ‘AI’ we’re seeing now isn’t actually new tech.
Huh, what do you mean with it’s getting that? I already have that.
Comma splice, dude.
Playing catch-up with Chrome again?
Does Chrome’s run locally on the machine, or does it ferry it over to Google Translate?
Firefox’s is done locally, it is not cloud based.
Also, I think it was already possible to use Google Translate and others through the extensions. This is an improvement because it’s done locally
Is there confirmation that it’s done locally? That would require a quite big download, especially if multiple languages are in it. Also if local, it should be easier to translate large documents instead of emphasizing snippets. One always has to be suspicious of Mozilla these days.
One always has to be suspicious of Mozilla these days.
As far as I know Mozilla never lied on technical details, at worst they didn’t know about wrongdoings by their partners and acted once known about.
The translation being done locally is advertised and should be simple to check with it being open-source. It also has to download for a while on first use, as well as translations take a considerable amount of time depending on the hardware. E.g. for larger sites it takes a short while until the last paragraph shows up as translated.
It uses the Marian library via WASM (their wrapper for this is here) to do translations, which AFAICT is “AI” based (which I presume knocks the file size down quite a bit) - additionally, the language packs (I’m not sure what term to use here so I’ll just go with that) are not all bundled with Firefox, they’re downloaded when you first use them.
The previous incarnation of it, the Firefox extension’s repo was found over here - I assume the code is now within Firefox’s main repo since its built into Firefox now.
I think the way to know is to unplug the PC from internet and see if Firefox can translate stuff.
Only their word until someone does it with a sniffer. E: I suppose, or looks the source but someone answered better now.
Note: Unlike other browsers that rely on cloud services, Firefox keeps your data safe on your device. There’s no privacy risk of sending text to third parties for analysis because translation happens on your device, not externally.
Or just try it with no internet
Assuming it’s not cached and sent next time it talks, of course.
