[stbullard]

Also, I'd like to see the inverse of it: a model trained only on the last few weeks - or even minutes - of published information, or only on scientific papers published in the last year or two. Should present some interesting psychoses!

[sigmoid10]

That would be much harder to build, because current information will often still reference or mention previous information. Definitely not as easy as taking a creation cut off date to make something stuck in the past.

[dnautics]

as per anthropic, only about 250 documents of any given concept required.