To preserve a copy of the humanity's recorded knowledge, you'd have to keep a copy of the library of Congress + archives of all scientific journals + arxiv.org, and equivalents of it for other languages, like Chinese, Japanese, German, French, Russian, Spanish, Hindi,.. Then there is a ton of proprietary and sometimes secret information held by companies and crucial for their functioning.
Where would you find this?
/s
Include a key logger and a daemon to grab full screenshots every few seconds.
Instead of just emailing everyone inviting them to join and telling them which kind friend reccomended they join - use the birthday and home address information and send them real tangible marketing material for their birthday (+ on Anniversaries, Christmas, Easter and on any day ending with a "y").
Choose some reasonably good llm model + corpus of data to enhance the generation (e.g. wikipedia).
Package it in a long-lasting battery, rough device. Think of a tablet/laptop you would take deep into the desert for a multi-week trip.
replace education in places without electricity with survival in case of civilization end.
Like, "explains it as well as a person who doesn't know anything about it but is reading the wikipedia page".
Like, "explains it but lies, and when you catch them at it, insists that they weren't lying".
Like, "can't actually do math, but has had heard lots of math problems so they guess and hope you don't check on them".
A sleazy marketer's idea of "explains it to you".
A lot of historical information on the Internet only scrape the top 50% of the knowledge on niche subjects, if even that. So often I see forum requests along the lines of "can someone scan page 242 of book X please".
https://www.wired.com/2017/07/inside-cubas-diy-internet-revo...