DAE notice AI tools got way better at reading old handwritten records after that genealogy conference in SLC?
I went to a small workshop in Salt Lake City back in March and a guy demoed an AI model that read 1800s census handwriting and got 9 out of 10 lines right. Before that I thought OCR on old records was hopeless, we were still typing everything by hand at my library branch. We ran it on a stack of county marriage records from 1870 and it pulled names and dates in about 20 minutes instead of a full afternoon. Now I keep wondering what other old paper piles I've been avoiding. Has anyone tried AI on church records or land deeds and had it actually hold up?