I agree with the overall premise, that it will be truly overwhelming for future historians to sift through the terabytes of historical information we are now accumulating. Even the US President's current output on Twitter and in speeches, debates, interviews - it's a massive corpus of information that the average human is hopeless to wrap their memory around. Furthermore, there is also the issue of whether or not data is locked up in servers, like Gmail, whereas previously letters or diaries would be open to family/biographers after the fact.
"It could be that future researchers will be required to form hypotheses first, which they then must test for explanatory power against the “historical record” that only AIs can actually query in its entirety. This would be a remarkable flip of the traditional model of historical inquiry, in which researchers use primary sources to then form explanatory theses."
This is great in theory, but I think this is an optimistic view of how history is done. Humans are inherently biased, and historians very often fall into ideological camps and approach questions with preconceived notions about what the answers might be. Why should historians be immune to these effects, when we know from sociological/psychological research into the scientific method that researchers do indeed have their interpretations colored by their a priori beliefs.
"It could be that future researchers will be required to form hypotheses first, which they then must test for explanatory power against the “historical record” that only AIs can actually query in its entirety. This would be a remarkable flip of the traditional model of historical inquiry, in which researchers use primary sources to then form explanatory theses."
This is great in theory, but I think this is an optimistic view of how history is done. Humans are inherently biased, and historians very often fall into ideological camps and approach questions with preconceived notions about what the answers might be. Why should historians be immune to these effects, when we know from sociological/psychological research into the scientific method that researchers do indeed have their interpretations colored by their a priori beliefs.