I don’t understand how they are supposed to “sell your data” if you just never use a Mozilla account and uncheck all the telemetry. Its not like they can secretly steal your data, since its Open Source.
It seems to me like just more FUD that Google is spreading to undermine our trust in free software.
There is no ai in ai, it’s just a program. In case of firefox translations it was made specifically for machine translation and later adapted to use with firefox browser.
well duh. But why is Mozilla investing in LLM research then?
Firefox translate isn’t an llm afaik. Anyway, why wouldn’t they? LLMs aren’t inherently good or bad, they are useful, and pushing for offline local first ai is a good thing.
is the LLM or its trainingdata opensource though, do we know what it does/is trained on?
There are truly open llm, there are somewhat open. If you are asking about firefox translate, it seems that models themselves weren’t trained by mozilla.