Heads Up -- FamilySearch Record Hints may Produce the Wrong Document Image

It may be difficult to read from the screenshot, but this particular record hand for Xaverio Corsi GQHR-PGR is an example where the image of the record given did not contain his name or entry at all. However, it was possible to find Xaverio by manually searching the records in the collection, one-by-one. 

This is another example of a record hint that was inaccurate. 

Here is a comment from one of my friends about the situation. 

When I check the hints in my Italian tree, The images associated with the hints are never correct. If the system gives me 3 or 4 hints that are duplicates, the 3 or 4 images will all be different and wrong. Most of the time the correct image is in the same book and I can find it with the date but not thanks to AI. For Pellegrinas' son Raffaello, his baptism was on 4th of November 1834, one hints showed me an image from 1832, the other  from 1843 and another from 1871. None for the correct date. When I asked AI about the discrepancy, this is what I got: This link, 3 times offered, shows 3 different pages for the document. It should be 1834, it shows 1832, 1843, 1871...

It should be noted that FamilySearch Record Hints are typically generated through a mix of automated record indexing, traditional automated record-matching algorithms, and probabilistic linking rather than generative AI. Also, AI cannot search the records directly and can only see a record shown on the screen. 

The fact that the Record Hints may not show the corresponding digital record should be a warning to those who use the feature to check the image record. It may also be the case that these Italian records have not been verified.