One of Benjamin Franklin’s most compelling quotations (and that’s a high bar) comes from his 1738 Poor Richard’s Almanack: “If you would not be forgotten as soon as you are dead and rotten, either write things worth reading, or do things worth writing.” In theory, this advice is sound -- and it’s actually even more applicable today than it was in Franklin’s time. The gap between then and now reflects the fact that it was far easier then to get things done and to be written about if one came from Franklin’s demographic than if one were a woman or a person from non-English stock.
■ A lot of stories are missing from most of human history, because they would have been written by or about people who weren’t in the most privileged classes of people. This deprives us today, both of the knowledge of their stories and of the context we should have about their times. If history had been written only by the left-handed or by people with hazel-colored eyes, something would be missing -- the problem is, we wouldn’t really know what. A great deal is missing from our history, and we can only try to claw back a tiny fraction of what that is.
■ This deficit in our knowledge has been problematic for generations, but now we face a truly bizarre turn of events: Generative AI tools make it possible to put words into the mouths of people from the past. “George Washington” can call himself “Bro-mander in Chief” and “Albert Einstein” can “teach” math class.
■ There’s an extremely narrow pathway via which versions of some missing stories could be told through careful reconstruction of the evidence. Down this particular pathway, some of the missing stories could be told -- though they would still need to be carefully labeled as reconstructions based upon limited available evidence.
■ Down another pathway, some of these stories could be told, but badly. Instead of evidence-based reconstructions, we could get hallucinations or confabulations -- the kind that has already prompted dangerous errors in the real world, with grave potential consequences. This could end up contaminating the real historical record if unaccountable knockoffs of important stories like the Narrative of Sojourner Truth find their way into competitive circulation.
■ Most likely of all, though, is the possibility that chat experiences with “George Washington” and other already-prominent figures will crowd out interest in other subjects. It’s a “rich get richer” problem: Those who are already prominent as historical figures left behind the largest records, and those records will make them more attractive as subjects for further content generation. AI “Thomas Jefferson” will probably attract a lot more attention than AI “Sally Hemings” -- even though their stories are conclusively intertwined in the genetic record.
■ Gaps in the historical record are more than just curiosities. They tell us where our understanding of both the past and the present are subject to imbalances. Those imbalances are worth examination and correction, but we have to make profoundly careful and thoughtful choices about how to rectify them.