For the past year, I’ve been working on a manuscript. Early in the process, I’d done some research to determine if Person A was related to Person B. They were in the same location at the same time, and shared a last name. It was a reasonable question to pursue and address.
Digging showed they were not related, so I moved on without including that tidbit in the manuscript.
Recently, I had a group of Early Readers review the manuscript for basic content: was it interesting, were there things missing, was anything confusing, etc.
One of those early readers asked if Person A and Person B were related.
I looked for my research notes. I am usually meticulous in documenting my research, in case it is needed again. Yet…I had not kept any notes. If one of the first people to read the manuscript asked, odds are good that others would ask, too. So, it needed to be addressed in the manuscript.
For due diligence, I did a quick online search to confirm that they were not related.
However, the newly-rolled-out feature that presented an “AI-assisted summary answer” at the top of the page presented different information than what I’d originally found.
This AI-generated answer rather enthusiastically said that Person A and Person B were, in fact, related. First cousins in fact! Their fathers, along with the names of those men, were brothers. What?!?
I was surprised and confused that I’d not uncovered such a simple thing in my initial research.
So, I began from the beginning. Again.
After about six hours of my own time as well as several hours from two friends who have jobs at area museums, both with access to sophisticated and complicated genealogy programs. Between the three of us, we were able to confirm my initial research. Person A and Person B are not, in fact, related…at least not as far back as the 1600s, at which point the trail goes dark.
Why I Rely on Primary Sources
That frustrating diversion is exactly why I tend to work extensively with Primary Sources. I might use secondary sources to find a path to follow, but I will not include in my own writing anything that I cannot confirm with at least one, and preferably two or three, primary sources.
That insistence has allowed me to feel confident when answering questions or musing on something. If there is something that is not known, or that we cannot ever know, I will word it as such (and in the footnote to the Person A and Person B situation above, I include that they could perhaps share an ancestor before 1600).
I’ve read secondary sources to find a path, follow the path, then find that something had been transcribed incorrectly and published with a wrong date or name or other detail. I’ve also found that my own interpretation of a statement in a letter or diary may be quite different than another researcher’s take on it.
I have developed the confidence to step forward and say that I have a different translation, and why.
But it all begins with the original documents, not a secondary source.
So, I have concerns about how research will be affected if others are not as rigid in using original source documents (which, sometimes, is itself inaccurate).
It’s also why I usually thoroughly document where I found an original document, save a copy of it, have a citation listing ready for use, etc. Usually. Had I done that for the Person A and Person B situation, I’d have not ended up down that dead-end rabbit hole caused by a hallucinogenic “helper tool” that is being thrust upon us now with every simple search.
For this one, I’d gone to the census records for three separate decades, but also wanted confirmation from the genealogy databases. Now I feel confident in my answer.
It’s also a good reminder that despite what the AI summary says so confidently at the top of a search page, it may have made the entire thing up.
Ugh.