Here's a *weird* thought:
There's a form of "care of the dead" through interacting with LLMs.
Imagine - data trained on all of history's knowledge, historical entries carry some more weight because they've persisted over time. LLMs output text from the past. It acts as sort of model for what these writers (with the entities that engineered it) want.
You might not get an exact copy of the text, or even what the authors exactly mean, or what secondary sources mean in interpretation, but that desire is essentially coming from the past. A sort of encouragement for us to understand in a different way, possibly even from their perspective.
