I predict that, in a couple of years, a benefit of knowing a non-English language will be to search for information written by humans. The LLMs are only fluent in English.
@arborelia
2050: Lojban because it's logical.
2060: Ithkuil, because it's Turing complete. And if it's not, it should be.
@arborelia the year is 2039. it is 3 AM and i am furiously studying Sentinelese. if i can become fluent by 7, then I'll have a good 18 minutes before the chatbots catch up and take up all the good search results. i'm ahead of schedule. i take a sip from my water bottle, the one i got so that i could study Xhosa, eyeing the time on the wall. the clock hands taunt me with their movements, jittery and rigid. my pen meets paper again and i continue my linguistic manifesto.
@arborelia: Not true. They'll be fluent in other major languages soon thereafter. Probably Indo-European languages first, so something like Japanese or Tibetan might take a bit more time.
But there's still thousands of small languages, which electric parrots will be sucking at for probably some 10–15 more years. So, if you genuinely want to learn a new language so as to exclude machines, consider small languages like Finnish or Basque, or languages that rich Europeans have unfairly assigned low prestige such as ǃKung or Dhuwal, or even, let's say, a language that humans gave up at some point, and would need to rebuild for use in modern world, like Aramaic or Gothic.
@arborelia this is resonating with me, being in the middle of reading Kuang's Babel https://en.m.wikipedia.org/wiki/Babel,_or_the_Necessity_of_Violence
2000: you search in English because there isn’t enough on the Web in other languages
2010: you search in English because what you find in other languages is SEO scam “translations” by WebHostingGeeks dot com
2020: you search in English so you can find Reddit results, the ones that still help
2025: you search in French because the chat-bots can’t fake expertise in it
2030: same but in Indonesian this time