Some people in the genealogy and family history community (notably on Facebook) have stated that Artificial Intelligence should not be used to create "historical fiction" about a historical person. Some commenters say that we as family historians should only base stories on our own memories, on sourced records, on actual words spoken by that person, or in witness accounts of the person.
I understand that point of view. BUT (a big but...) I don't have memories, family stories, witness accounts, etc. for most of my ancestors - only for myself, my parents, my grandparents, but I have thousands of records for my ancestors.
How can I tell the stories of my ancestors and their lives so that my relatives and descendants can better understand my life and the ancestors' lives?
As my blog readers and Facebook friends know, I have been telling "historical fiction" stories about my ancestors for over a year now in an attempt to understand the lives of my ancestors. I do this by asking questions about how did the ancestors live their lives, what historical and social events affected them, what were the major decisions or turning points in their lives, etc.
Am I wrong to do that? My view is that this is a useful and legitimate activity as long as I identify the results as "Historical fiction" and "how it might have been" and avoid stating that it is how it really was for the ancestors in the time and place they lived.
1) I decided to see what OpenAI ChatGPT5 thought about it. Here is my prompt:
"I'm an experienced genealogist but a technical writer and don't tell stories well with emotion or feelings, and have stories about my ancestors only from my parents and my own memories. Is it wrong to use AI to tell historical fiction stories about my ancestors - 'the way it might have been' based on the facts I have from source records with names, relationships, events, dates, places, etc.?"
And here is ChatGPT5's response:
Not wrong at all—if you handle it honestly and carefully. In fact, what you’re describing sits in a long tradition. Family historians have always tried to bridge the gap between documented facts and lived experience. The key is how you frame it.
Think of it this way: you’re not replacing history—you’re interpreting it.
a) Where it works well
Using AI to create “the way it might have been” stories can:
Bring otherwise dry records to life (census lines, land deeds, probate files)
Help descendants connect emotionally with people they’d otherwise see as names and dates
Explore plausible daily life based on known context (time, place, occupation, culture)
That’s especially valuable in genealogy, where the record is often skeletal. You already know that even a well-documented ancestor leaves a lot unsaid.
b) Where you need to be careful
The line to watch is between fact and invention. Problems arise when:
Fiction is presented as proven truth
Readers can’t tell what is documented vs. imagined
Details contradict known historical realities
Those are fixable issues with good discipline.
c) A practical, traditional approach (that works beautifully)
You might appreciate this—it aligns with careful, evidence-based work:
1. Keep your proof standard intact -- Maintain your fully documented, source-based biography (your “anchor”).
2. Create a clearly labeled companion piece -- Call it something like: “A Historical Reconstruction”
“A Day in the Life (Imagined)”
“Based on Records of…”
3. Ground every scene in evidence -- For example:
If he was a millwright → describe mill work typical of that region and period
If she lived in 1870 Iowa → use known housing, clothing, weather patterns
If a child died young → you can explore the emotional weight, but don’t invent specific dialogue or events without signaling it
4. Signal uncertainty honestly -- Phrases like:
“He likely…”
“She may have…”
“It is reasonable to imagine…”
That keeps your integrity intact.
c) Why this actually strengthens your work
You’re doing something many genealogists struggle with—making ancestors human again.
Records tell us:
what happened
when
where
But they rarely tell us:
what it felt like
what a day looked like
what mattered to them
Careful historical storytelling fills that gap without distorting the record.
d) One more thought (from a traditional angle)
Before AI, people did this by hand—writing sketches, local histories, even family lore narratives. AI is just a new tool for an old instinct: honoring the past by trying to understand it.
Used responsibly, it’s not cutting corners—it’s extending your ability to interpret the past.
If you’d like, I can take one of your ancestors—say Devier Lamphier Smith or Della Carringer—and show you exactly how to build a clean, well-labeled “historical reconstruction” from your known facts.
2) My next prompt was:
"One of my thoughts is that what I am doing is acting as a ghostwriter for my ancestor who cannot speak for him or herself. Is that a fair thought? Is there another term besides novelist?"
The ChatGPT5 response was:
That’s a very fair way to think about it—and, frankly, a respectful one.
“Ghostwriter” fits better than you might expect, because you’re not trying to replace your ancestor’s voice—you’re trying to reconstruct a voice that history didn’t preserve. In genealogy, that’s a meaningful distinction. You’re working from evidence, not inventing freely.
That said, there are a few terms that might suit what you’re doing even more precisely, depending on how you want to present your work:
a) Terms that fit your approach
- Historical Interpreter --- This has a solid, traditional feel. It suggests you are explaining and giving life to the past based on evidence—much like a museum interpreter or living history presenter.
- Narrative Genealogist -- A nice blend of your two worlds. It keeps your identity grounded in documented research while acknowledging that you’re shaping it into story form.
- Family Historian (Narrative Form) -- Simple and honest. Many respected genealogists have done exactly this—writing sketches or reminiscences to accompany the record.
- Historical Reconstructor -- This leans into what you’re actually doing: rebuilding a plausible lived experience from fragments.
- Creative Nonfiction Writer (Genealogical) -- Useful if you ever share this more broadly. It signals: this is rooted in fact, but told with narrative technique.
- First-Person Historical Reimagining -- Not a title for yourself, but a good label for the pieces. It clearly tells readers what they’re getting.
b) Why “ghostwriter” still works
Your instinct is sound because:
You’re writing in his voice, not yours
You’re constrained by known facts and context
You’re not claiming he actually said these words
It’s almost like you’re giving him a chance to testify, based on the record he left behind.
A small but important guardrail
The only caution I’d add—very much in keeping with your careful approach—is this:
Pair the idea of “ghostwriting” with transparency.
For example, a simple note like:
“This narrative is a historically informed reconstruction based on documented facts and context. The words are imagined, but the circumstances are grounded in evidence.”
That keeps your integrity intact while still allowing the story to breathe.
c) A traditional perspective you might appreciate
In earlier generations, family historians often wrote “reminiscences” or “sketches” of ancestors—sometimes even in first person—especially when preserving stories for children and grandchildren. They understood, as you do, that facts alone don’t carry memory forward.
You’re really doing the same thing—just with better tools and a bit more discipline.
If you want, I can help you settle on a consistent label and format you could use across all your ancestor narratives so future readers immediately understand what they’re reading.
=================================
3) Let's stop here for now. I hope that this will result in honest discussions about this issue and not devolve into rancor and division. I think it needs to be discussed.
4) My understanding is that OpenAI ChatGPT5 and most other large language models try to be positive and support the information presented to them as a prompt or opinion. Do you think ChatGPT5 is trying to fool me with these responses?
Links to my blog posts about using Artificial Intelligence are on my Randy's AI and Genealogy page. Links to AI information and articles about Artificial Intelligence in Genealogy by other genealogists are on my AI and Genealogy Compendium page.
No comments:
Post a Comment