Before You Ask AI, Ask Your Nana
One of my cousins called me last week to pick my brain about our family tree. I do this professionally now, within the Heritage & Legacy services of Vysyons Communication, but before that, I did it with and for my Nana. I've been at this since 1985, when she started teaching me our family tree, and have branched out into what I've come to think of as my very own genealogical forest.
When he called, I expected the conversation many genealogists have with family members all the time. Who is this? When did this happen? Where were they born, died or buried? Instead, I was surprised when I heard that he wanted me to clear up what he had learned in his research, on an AI app. He didn’t dive into Ancestry.com or FamilySearch.org. He talked with an elder briefly. He decided to ask questions, without
I want to be precise here, because I don't think he did anything wrong. I think he did what feels natural nowadays, which is to ask one of the tools that everyone uses on an elementary level to get answers from. The thing is that family history research isn't like most other questions, and the answers need to not be as well. Also, the order you research matters as much as the research itself.
You have to start with the people who lived it. Elders and family members who hold the history are the only source to the roots of your family that make everything else make sense. There knowledge can give context that no document will ever contain. They know the reason why the family lived where they lived, and why they moved; the secret to the mysteries that have only been whispered and hinted about; and the name that never made it onto any form. Most times that context can only be found in a person for exactly as long as that person is here to tell it. And with each year, and passing, access to that context lessens more and more. There is no backup.
Then go to what's written down. First should be the family documents, such as letters, Bibles, and photographs with names on the back, and hopefully with older family members to continue providing context. Next, the archival record, such as, census records, birth, marriage and death certificates, and newspaper archives, this is the paper trail that exists specifically because churches, governments, etc. were required to write your family’s information down. This is slow work because not everything is digitized, or information was not accurately or complete documented. Even worse is that some documents no longer exist due to a variety or reasons. This information gives you you proof, documents strengthen the stories.
DNA comes last, and it comes last on purpose. It's extraordinary for what it does well, clarifying an adoption, answering a biological question no document can settle, connecting you to a line the paper trail ran cold on. It answers a narrower question than most people expect, which is exactly why it belongs after the first three steps have done their work.
So where does a chatbot fit into that order? Honestly, near the very end, if at all. And only after the real research is done. After talking to my cousin, and looking over what he got from his AI search, I decided to put it to the test before writing this. I used the same AI platform he did to look into four of my own documented ancestors. These were people I had birth and death dates and locations, but had questions of I was stuck on. It came back with nothing at all at first, and then limited answers as I provided it more and more details about these people. This was not because the tool is broken, but because the records that actually answer genealogical questions live behind logins, subscriptions, and archive doors that open web search simply cannot reach. Without that access the chatbot cannot generate a family tree, and can only provide vague answers. That is the actual danger. A wrong answer and a right one look identical on the page, especially if you have no context or documents to start with
In my opinion, I think AI belongs in creating outlines, providing names of suggested resources, and organizing the gathered stories and facts coherently. AI cannot do the interpretive work that makes genealogy, genealogy. It will not know which records to trust when two contradict each other. it cannot understand what a guardianship filing meant for a Black family in 1905. It cannot read the silence in a document as carefully as the words. It’s taken me forty pre-computer, pre-Ancestry.com, pre-DNA testing years to learn how to do all of this. This why people hire a genealogist like myself.
I told him what I'm telling you now, if you choose to do this work yourself, or eventually decided to hire a genealogist. Start with the people who remember. Move to what was written down. Let biology answer what only biology can answer. And if you choose to bring AI in along the way, let it organize what you've already found, not invent what you haven't.
Worth Noting:
If you're sitting on a family story you've never been able to place in a document, that's usually where the real research begins, not where it ends. That gap is often the most interesting thread in the whole tree.