Schmidt Sciences' HAVI extends Brent Seales's AI-decipherment lineage from carbonized scrolls to illegible shorthand — different obstacle, same underlying move

Brent Seales spent roughly twenty years developing "virtual unwrapping" — reconstructing a manuscript's text from a volumetric scan rather than physically opening it — to the point that it worked on the 2015 En-Gedi Scroll and, later, a fully carbonized, never-opened Herculaneum scroll. He now leads Schmidt Sciences' Humanities and AI Virtual Institute (HAVI), the institute that in December 2025 awarded $11M across 23 teams — among them Christlein and Weichselbaumer's project training AI to read Gabelsberger shorthand, the notation Kurt Gödel's own private notebooks are written in.

The bridge is institutional, not metaphorical: the same person and the same funding body that grew out of proving a burnt scroll's insides didn't need to be physically opened to be read is now backing the tool that would make illegible 20th-century handwriting legible to machines. The obstacle changes — carbonized fiber versus a dead notation system — but the move does not: AI deployed as prosthesis for a human literacy that no longer exists in enough living readers to matter.