Using AI to Transcribe and Translate Documents

Being able to upload an image of a document to AI and have it instantly and accurately transcribe the difficult to read handwriting or translate a foreign language from an equally difficult to read handwriting is an exciting idea, but in practice it does not work well. Specifically, the biggest problem I find with it is that when it can't read something, it decides to just... make something up that seems in keeping with the type of document it is. Clearly, AI is smart enough to recognize what kind of document it is and supply the type of information you'd expect from that type of record, but it is apparently not smart enough to realize why the information needs to be accurate and not fabricated.

Take for example this will of Henry Rorer (above, click to enlarge). I uploaded it to ChatGPT and asked it to transcribe it. This is what it gave me:

In the name of God Amen the twenty-second Day of January in the year of our Lord one thousand seven hundred and eighty. I Henry Rorer of Bristol Township in the County of Philadelphia Yeoman being weak in body but of perfect mind and memory thanks be given unto God, therefore this my last Will and Testament. That is to say, I first and principally I resign my soul to God that gave it and my body to the Earth to be buried in a decent and Christian like manner at the discretion of my Executors herein after named and touching such worldly Estate wherewith it hath pleased God to bless me in this life. I give and dispose of the same in the following manner, and form.

Imprimis. I give and bequeath unto Mary Willard one Chest that she is to choose out of all my Chests and Boxes to be paid and delivered her by my Executors hereafter named.

Item. I give and bequeath unto Margaret my dearly beloved Wife all my Estate both real and personal during her widowhood, excepting such legacies as I shall hereafter mention in this my last will and testament and if in case my Wife should die or marry, then it is my will that my Estate shall be sold and divided equally amongst my children or the survivors of them share and share alike, and if either of my Children should die under age or without lawful issue that then my Estate shall be equally divided amongst the survivors of them, and I do hereby nominate and appoint my beloved Wife and my Friend Henry Rorer and George Rorer Executors of this my last Will and Testament, and I do hereby utterly revoke and disannul all former Wills and Testaments by me heretofore made Ratifying and Confirming this and no other to be my last Will and Testament.

In witness whereof I have hereunto set my hand and Seal the day and year above written. Signed Sealed published pronounced and declared by the said Henry Rorer as his last Will and Testament in the presence of us.

Phillip Hagey
Peter Shall

Henry Rorer (Seal)

At first glance, it looks great. But when you actually compare it to the original document, it quickly becomes clear some of it was just completely made up. Firstly, note that the date is slightly off. It was actually the 29th day of January, not the 22nd. It's understandable a single word might be misread, but it goes downhill from there. Where the transcription reads "thanks be given unto God, therefore this my last Will and Testament" the record actually says "do make and ordain this my last Will and Testament." The transcription then goes on to add in an entire sentence that isn't included in the record: "That is to say, I first and principally I resign my soul to God that gave it and my body to the Earth to be buried in a decent and Christian like manner at the discretion of my Executors herein after named..."

Although not accurate, it's at least not giving us any misleading information. The paragraph following it, however, is fabricating people and property never mentioned in the original. The transcription says:

Imprimis. I give and bequeath unto Mary Willard one Chest that she is to choose out of all my Chests and Boxes to be paid and delivered her by my Executors hereafter named.

The actual document says:

Imprimis It is my Will and I do order that in the first place all my just Debts & funeral charges be paid and satisfied.

ChatGPT did not even attempt to get it right. It's as though it ignored even the words it could make out in favor of a sentence that makes sense instead of just leaving blank spaces where it isn't able to read certain words. So the entire sentence was changed just to produce something that wasn't missing words. This could be really disastrous to your research if you're not paying attention. It fabricated a name where there was no person mentioned, making it look like Henry Rorer willed one Mary Willard her choice of a chest or box when there was no such person ever mentioned in his will and no mention of any chests or boxes either.

Additionally, the transcription looks surprisingly short. Even though the will is only about one page length, it seems like more than what the transcription included, and indeed, it is missing a lot of information from the rest of the document including many people who are of importance to Henry Rorer.

The same thing essentially happens when you ask ChatGPT to translate a foreign handwritten document, wherever it struggles to read the handwriting, including names, occupations, ages, etc, it simply makes up random data to replace it. It changes names and other data just to produce a complete transcription, whether accurate or not.

So here is your warning. AI is not there yet. Maybe there's more reliable ones out there, but I imagine they aren't free. Whatever FamilySearch is using to transcribe records in their experimental Full Text Search using some kind of OCR for handwriting seems to work much better than ChatGPT. But not all records are on FamilySearch, let alone under their Full Text Search.

As ever, the genealogist mantra remains true: you have to do the work yourself and verify everything.

When it all comes together

Mary Cath. Brady, Martha Washington House, William Henry
McBride, and Daniel H McBride are all children of the John
McBride in my DNA match's tree, mentioned here as the heirs
of Catherine McBride, the mother of "my" John McBride.

This is a great example of how a combination of DNA and paper research can break down a brick wall. And it's why you should read all pages of a multipage document, even if it doesn't seem like it has any useful information at first. Additionally, probates often seem to get overlooked, but this is also a great example of how important they can be. When you're stuck, always look for a probate record, of all relatives involved. Even if they are female (women sometimes had wills and probates too!).

I had a suspicion that one of my DNA match's ancestor was my 5th great uncle, John McBride. The DNA match had the same name in her tree, but she knew nothing about him apart from his name (which she found from orphan court records), his wife, and his children. I had his birth and death data, and obviously his parents names (and records to back it all up), but no records confirming his wife's name or any children. The only thing they had in common was both location and time period, but there was no proof they were definitely the same man. It was like I had half the story, and she had the other half, but we had no way to link them together.

I finally read ALL the pages of John's mother's probate records. Initially I'd only read her will, thinking if she names her grandchildren by her son John, and they match the names of John's children in my DNA match's tree, that obviously proves they are the same man. The will doesn't name them (only says "my grandchildren by my son John")... but upon further inspection of the follow up documents, such as the distribution of her estate, it does list several people whose names match perfectly with the children in my DNA match's tree! Although it doesn't specify they are her grandchildren, given the context (i.e., her estate is being distributed to her heirs, as specified in her will as her grandchildren), it would be too much of a coincidence for so many of them to be listed on this woman's probate records if they weren't her grandchildren.

So when you're struggling to find a connection to a DNA match, it pays to do some digging around on a hunch, even if it seems like a long shot or there's not enough info to say for sure. The only connection I had was a name, and an extremely common first name at that, with a surname that isn't unheard of either. Even my DNA match in question was skeptical when I first proposed the idea to her, but a little digging proved my hunch was right!

Things You Should Know About The New Probate Records at Ancestry.com

Christian County, Kentucky Will Book Q, p. 172-173

Ancestry.com recently added a boat load of probate records from many, many counties in many, many US states. The best part is, that they are digitally indexed so unlikely many of the probate records at FamilySearch.org, you can do a name search to find them! The overwhelming amount of them being added all at once was very exciting, but after looking into it, here's some realities of the collections:

1. Not all records are digitally indexed. I was having difficulty finding anything so I tested it on a will I already had - Leonard Wood in Christian County, Kentucky 1857 (will book Q, p. 173). I couldn't find it with a name search, but did find it by looking manually. It's not a case of it being transcribed wrong, it looks like it's just not indexed at all (the button to bring up the digital index is grayed out on both the will page and the handwritten index pages). I can't even attach to my tree - the only save options are save as web link or save to my computer. Lots of people are going to miss a lot of things when this appears to only be half-indexed.

2. Many of the records were already available on FamilySearch.org. Granted, some of the records were not on FamilySearch, and the benefit Ancestry.com has over FamilySearch is that many of the records are indexed. But as addressed above, not all of them are.

3. Not all collections are complete. For example, the Philadelphia Wills begin in 1838 so anything before that is still not available online (FamilySearch.org doesn't have any Philadelphia probates). They do have earlier years for Administration records though. The date range for each county is difficult to see upfront because the collection is for the whole state, which is then broken down into counties. The specific date range for each county varies and isn't mentioned so you have to look through them to figure it out. Also, not all collections include all counties for that state. For example, the Pennsylvania Probates says it has records for about 90% of PA counties.

4. Not all collections have a manual index. For some, the odd missing digital index isn't a big deal, as many of the collections have a handwritten index in the images which you can manually browse. Unfortunately, not all of them do. The Philadelphia Wills, for example, do not.

Why Probate Records Are So Important.

Today, I made a remarkable discovery. Well, it's remarkable to me. It was accomplished almost entirely with the Pennsylvania Probate Records found at FamilySearch.org and is a testament to how important these records are and how much you can learn from them if you take the time to find and study them. It also proves research before the almighty 1850 US Census can be done.

Ann Sutch Will 1827 mentioning brother Richard
Shoemaker.

I had been searching for the parents of my ancestor, Ann Shoemaker, for a while. All I knew of Ann was that she married Daniel Sutch, had 4 daughters, and then died in 1827 in Gwynedd, Montgomery County, Pennsylvania. I did not even know when she was born but I approximated it around 1760. But I did also know she had a brother named Richard. I discovered this from her will in the Montgomery County Probate Records, proved in 1827, which specifically named her "brother Richard Shoemaker" as executor of her will (shown above right). This not only gave me her maiden name but also a brother's name to research. It was difficult though, because all I knew of Richard was that he was probably alive (and an adult) in 1827, and likely lived in Montgomery County. But Shoemaker was a common name in that area and Richard was not uncommon either. Without knowing anything else about him, how could I confirm records to be the Richard I was looking for?

Well, in the Proceedings Index for Ann (Shoemaker) Sutch, there were some listings for Orphan Court Dockets. These are often records that have something to do with the will of a deceased person after it was proved. There were two dated 1838 which turned out to be a petition and answer for the replacement of trustee Richard Shoemaker, deceased, with someone else. The petitioners were E. Jones and Job Roberts (I promise this will be important later on). This suggested that Richard Shoemaker, brother of Ann (Shoemaker) Sutch, died sometime in or soon before 1838. So I went looking for a Richard Shoemaker who may have had probate records dated around 1838. There was only one in Montgomery County who fit with this and although there was no will listed, there was an Admin Bond date for Aug 5, 1837 in Horsham and the Admins listed were Job Roberts and Evan Jones. So I knew I had the correct Richard Shoemaker because Jones and Roberts were listed in the Orphan's Court record for Ann Sutch, sister of Richard Shoemaker.

But that's not all. Once I entered Richard's death year as about 1837 in Horsham, a Quaker record on Ancestry.com popped up for a Richard Shoemaker who died July 10, 1837 in Montgomery County (subscription required to view this record). I looked at it and although it didn't say he died in Horsham (there was no death location at all), it did say his father was Ezekiel Shoemaker who had died 1816 in Horsham. I already had a hunch this was my Richard Shoemaker because in the Estate/Proceeding Indices, there was only one Richard Shoemaker who died in or around 1837 in Montgomery County (and if he died in July, a probate record in August made perfect sense). But just in case there was another one who perhaps didn't have any probate listings at all, I decided to research Ezekiel.

Firstly, I noticed on the Proceedings Index right above my Richard Shoemaker there was another entry for a Richard Shoemaker who died around 1790 in Horsham and his executor was named Ezekiel Shoemaker. I looked at his will first and sure enough, Ezekiel was his son. Best of all, two of his daughters married into the Roberts family, which linked this elder Richard and son Ezekiel back to my Richard, because if you recall Job Roberts was listed in my Richard's probate records (who would have been this elder Richard's grandson). Granted, Roberts is a common name too but there's starting to be too many coincidences to ignore. Additionally, according to other family trees, my Richard also married a Roberts.

Ezekiel Shoemaker 1816 Will naming his daughter,
Ann "Such" (Sutch).

I looked up Ezekiel in the probate records and fortunately, he had a will and sure enough, in his will he names "my daughter Ann Such" (shown left). So not only do I now have proof that Ann was the daughter of Ezekiel, I also already have Ezekiel's father's name as Richard, and Ezekiel's siblings names as mentioned in Richard's will! A wealth of information, with the exception of one record, came entirely from these probate records.

To top everything else off, I then found a Quaker death record for Ann Sutch who died 1827 naming her father as Ezekiel Shoemaker of Horsham (subscription required to view this record). These must be new records added to Ancestry.com since I'm sure I scourged the internet looking for another death record for Ann once I found her will and knew she died in or before 1827. My search would have been a hell of a lot easier if I had just found this record first! Regardless, I still would have gone in search of Ezekiel's will to find out more about their family (like his wife's name) so the point still stands that probate records are important.

For some reason, there is a secondary record with no indication of the source or repository attached to some Ancestry.com member trees that claims Ezekiel's daughter Ann "died young". I hope I have been able to conclusively prove that this is not true with all these primary records I've mentioned and provided links to. Family trees put Ann's birth year as 1764, not far off the estimated birth I made around 1760, so if this is true she would have been 63 years old when she died in 1827. She married Daniel Sutch and had four daughters named Jane (b. abt. 1788, m. Charles Gilbert), Sarah (b. abt. 1791, m. William Davis), Ann (b. abt. 1792, m. Homer Dubree), and Hannah (b. abt. 1805, m. Joseph Amber). Some information on their family can be found in the Ambler Gazette.

So don't overlook probate records as an important method for finding that elusive previous generation. It may take a lot of digging and it may not always lead back to what you're looking for but you will likely discover something you didn't know before.