> We will become reliant on a small group of trained translators and experts to report what history—including the documents and papers of our own families—was a

> We will become reliant on a small group of trained translators and experts to report what history—including the documents and papers of our own families—was about. I think we know this won't be true based on AI getting better and better at handwriting recognition, despite the downsides of offloading this work to machines (e.g. if trained in closed system, could result in censorship and potentially limited validation at the endpoint). However, I do see a lot of upsides to using computation for such efforts: increases the scale of processed documents by orders of magnitude and enables indexing, search, categorization, and further analyses which would otherwise take many man-months for trained translator and experts to complete.