Can we use AI to help us with genealogy?
The standard answer for the last couple of years has been, “Yes, but you’re going to need to play with it for 10-20 hours to really understand how to move forward.”
But maybe that’s not the case anymore. Newer versions of these products can think, access the internet, write code, and conduct research.
So last week at Chronicle Makers, when Denyse Allen offered up a new prompt designed to help AI newcomers, the timing seemed just right. Her promise? Use this prompt and avoid the 20 hours of blood, sweat, and tears.
You can see the full prompt at Chronicle Makers: The Worst Advice Genealogists Keep Getting.
I haven’t asked Denyse to outline her vision for the prompt, but my guess is she’s in favor of education and guidance. She just doesn’t see much point in trial and error.
The idea is to give it information about what you are researching and get back some pointers on where to focus your research next.
It starts by asking for:
Family lines you’re researching
Records you’ve already found
Sources you’ve consulted
Key findings and breakthroughs
Dead ends and challenges
The first thing you may be thinking is, wow, ok, this will take some time.
Well, yes! The good news is, genealogists know how to pull research together. We do this for fun on the weekends.
So hey, I’m game. Let’s give this a try.
Don’t Try to Feed it Your Whole Tree
Here’s where it makes sense to step back and think about goals. I’m not going to try to dump my whole family tree in here. In fact, to test this out and get accustomed to what it does, I’ll start with one person, my grand Uncle Orin Carver.
Orin was Grandpa Orland’s older brother. (Why name one boy Orin and another boy Orland? Didn’t that get confusing when it was time to call the boys to dinner?)
I don’t have much information on Orin, just one picture, a birth date, a couple of census sitings, and some rumors. Oh, and he often gets confused with another man named Orin Hugh Carver. There are some leads I haven’t followed, so we’ll see if ChatGPT picks up on those.
I’m particularly interested in a news article that says he was in an asylum for 18 months. The family is inclined to believe my grandfather had a mental illness, and there are hints of bipolar disorder in Great Grandma’s family, which leaves us curious. Sadly, the treatment options were not very advanced 100 years go.
I decided to compile my information directly out of Ancestry.com and then add a few notes. I didn’t worry about feeding in the source images, opting for transcripts instead.
I could have pasted this information into the chat box directly after the prompt, but I chose to attach it as a separate file. ChatGPT, Claude, Gemini, and Grok all have the ability to attach files.
Below you can see our Q&A and the plan it provided. Note that I have not yet attempted to follow the plan and cannot confirm whether all of the leads are correct. However, it seems like a solid starting point.
It picked up on the Washington death certificate, which I think is the logical next lead, and gave instructions for finding it.
If I were brand new to AI?
Now, how would this work for me if I were brand new to ChatGPT? Honestly, I would fumble a little bit. That’s just me.
I would probably want a couple of hours to play around, just so I could understand how ChatGPT works, the settings, the tools, how to attach files, when it makes sense to start a new chat, and so on.
But after that? I could use this prompt and get going.
The other thing to adjust to with these AI assistants is the idiosyncrasies. Each product has its own quirks, which change from model to model.
They may provide incorrect guidance, such as links to places that don’t exist.
Some days their output is less than useful, and your best bet is to start a new chat.
If you provide links to web pages they can’t read or can only partially read, they will sometimes pretend they actually read them and then try to fake their way through the conversation.
But there’s no reason you can’t use the new, focused prompt while you are learning to deal with these idiosyncrasies.
What if you have a lot more data?
If you have a lot more source material to attach than I did here, you can attach multiple files instead of putting all of your information in one file.
Behind the scenes, these files can be added as PDFs or they might be stored in your Google Drive or Microsoft OneDrive.
Another interesting option for paid subscribers would be the “Projects” feature in both ChatGPT and Claude, which lets you keep all the files associated with a particular project together in one place.
My thoughts on the prompt so far?
I’m a fan. Get your research together and give it a try. And remember, start small. You don’t need to dump your whole tree in there!
Here’s the conversation with ChatGPT
Let’s clarify your goal for this research:
Primary Goal: What specific outcome do you want? Is it confirming a death date, proving the asylum claim, separating him from Orin Hugh, or something else?
My answer: I am trying to trace his life in general, so please tell me which of these would be the easiest to start with.
Time Period of Focus: Are you most interested in the 1880s asylum period, the 1910s (California mention), or the 1925 possible death?
My answer; I am interested in them all. However I'm not terribly convinced that he was ever in California, any more than I'm convinced he was killed in a prairie fire or living in Gresham.
Geographic Focus: Are you focusing on California, Iowa, or Washington? Or just trying to follow wherever the trail leads?
My answer: I am trying to follow where the trail leads. I have seen no clue of him in California.
Family Line Priority: Are you researching Orin primarily because he’s a direct ancestor, or because he’s part of the broader Carver family puzzle?
My answer: He is part of the broader Carver family, my grand uncle.
Personal Interest: Is there a particular reason you want to pin down Orin’s story — family mystery, local history tie, potential descendants?
My answer: Because I don't know anything about him other than he was Uncle Nin. Interested in his medical background.
You can jot those out in a quick reply, and I’ll move on to the planning stage once I have that.
Why the 1925 Washington lead is the easiest opening move
The Washington Department of Health Death Index, 1907-1960 is fully searchable (and free) on the Washington State Digital Archives. It gives the certificate number you’ll need.
Washington now hosts scanned death certificates to 1996. If Orin’s certificate image isn’t online yet, you can order a non-certified copy for $25 from the DOH by mail or online. Turnaround is fast.
Eastern State Hospital retains historical patient files, and their records office answers genealogical requests.
A death certificate plus a patient file (if it survives) should pin down residence history, marital status, burial place, and—in many cases—the admitting hospital or court that sent him there. That can propel you backward into earlier censuses, voter rolls, or court commitments.
Step-by-Step Research Plan
Pull the 1925 death certificate
Use the Digital Archives index to grab the cert #, then request the image or order the copy.
Request Orin’s Eastern State Hospital patient file
Email him.esh@dshs.wa.gov with his full name, death date, and certificate number. Ask for admission, discharge, and burial data. Some files include next-of-kin correspondence.
Search Spokane-area newspapers July 1925
Look for an obituary or burial notice that might list surviving relatives or a transfer of remains.
Rule out California by confirming 1914 wording
Locate the exact 11 Jun 1914 Verdigre Eagle article on Newspapers.com or Chronicling America. If it simply says “in California,” treat it as unverified until census, city-directory, or voter-roll entries appear.
Re-examine the Independence, IA asylum lead
Contact the Independence Mental Health Institute records desk (IndeMHIMedicalRecords@hhs.iowa.gov). Ask whether admission registers 1885-1890 include an Orin/Orien Carver born 1858-60.
The File I Attached
Note that I’ve added XML tags around each piece of information (the brackets.) This isn’t really necessary, but it will allow me to refer to each record by name if I want to discuss it with ChatGPT. I can say things like “The birth date in <1880 Census> vs the birth date in <1870 Census>” and it will be very clear what I mean.
<Basic Information>
Name: Orin Carver
Known Family Nickname: Nin
Birth: 1859 in Wisconsin
Death: unknown
Father: Cyrus Hoyt Carver
Mother: Mary Jennette Allen
1870 - Residence La Valle, Sauk, Wisconsin, USA
1877 - Family moves to Verdigris, Antelope, Nebraska, USA
1884-1886 Perhaps in Insane Asylum in Independence, Buchanan, Iowa, USA
1891 - Death of mother Verdigris, Antelope, Nebraska, USA
1814 - Death of father Knoxville, Knox, Nebraska, USA
1914 - Alive, per father’s death notice in The Verdigre Eagle, Thu, Jun 11, 1914, Page 1.
1914 - “In California” per father’s death notice in paper
</Basic Information>
RECORDS I’VE FOUND
<1870 Census>
Name: Orin Carrer
Age in 1870: 12
Birth Date: abt 1858
Birthplace: Wisconsin
Dwelling Number: 51
Home in 1870: La Valle, Sauk, Wisconsin
Race: White
Gender: Male
Post Office: Baraboo
Inferred Father: Cyrus H Carrer
Inferred Mother: Mary J A Carrer
Household members
Cyrus H Carver age 52
Mary J A Carver age 42
Dorr H Carver age 17
Orin Carver age 12
Loren E Carver age 9
Oliver A Carver age 7
Donnelly Carver age 5
Barton B Carver age 2]
</1870 Census>
<1880 Census>
Name: Orin Carver
Age: 21
Birth Date: Abt 1859
Birthplace: Wisconsin
Home in 1880: Sherman, Antelope, Nebraska, USA
Dwelling Number: 13
Race: White
Gender: Male
Relation to Head of House: Son
Father's Name: S.h. Carver
Father's Birthplace: New York
Mother's Name: Mary Carver
Mother's Birthplace: New York
Occupation: Farmer
Attended School: Y
Household members
S.h. Carver age 70
Mary Carver age 59
D.h. Carver age 29
Orin Carver age 21
Loren Carver age 18
Oliver Carver age 16
Donaley Carver age 14
Barton Carver age 12
Orlen age 10
</1880 census>
<Family Lore>
“Died in prairie fire”
“Living in Gresham, Oregon”
</Family Lore>
<Neligh Leader>
Aug 1887 Neligh Leader: “Orin Carver son of Chas. Carver, is in the asylum for insane at Independence, Buchanan county, Iowa. We are informed he has been there for the last 18 months.”
Comment: Neligh is a bit south of the homesteads, but in the general area, and our Carver family is the only Carver family to ever be mentioned in the papers during this time frame in this area. So it is possible this is our Orin with his father’s name misreported, but no way to be certain.
</Neligh Leader>
<Washington Death Index>
Note: Not proven to be him.
Comment: His brother lived here during this time frame so maybe this is him, but who knows.
Washington, U.S., Select Death Index, 1907-1960
Birth, Marriage & Death
Quick compare
Name Orin Carver
Birth abt 1858
Death 30 Jun 1925 Medical Lake, Spokane County, Washington New
Burial 01 Jul 1925
</Washington Death Index>
SOURCES CONSULTED
Ancestry.com
Familysearch.org
newspapers.com
nebraska historical society
KEY FINDINGS AND BREAKTHROUGHS
Key finding: Many online trees confuse this Orin with another named Orin Hugh Carver and mix them together.
Potential breakthroughs: The recent news article about the Insane Asylum is new and potentially interesting. Perhaps not actually a breakthrough.
DEAD ENDS OR CHALLENGES
I just don’t find much about him.
Love the consumer reports review of my prompt. I have to say it gave you some high quality sources rather than the usual drivel which was my goal in writing it. What I’ve observed over and over again is that when people play with AI, they do the equivalent of Google Searches and go off on a trail from there to an unpredictable place.
I think you need to get the death certificate. I was interested to see the prompt in action.