When One Question Leads to Another: How ChatGPT Changed My Genealogy Research

 

Have you ever started researching one question only to find yourself investigating something entirely different?

Genealogy research often works that way. We find one record, which raises a question, which sends us looking for another record. Before long, we're following a research trail we hadn't anticipated.

I've discovered that my conversations with ChatGPT can evolve in much the same way.

A recent conversation about James Marshall Ricketts is a good example. I started with a fairly simple request:

Create a timeline from the transcription of James Marshall Ricketts's Civil War pension file.

I wasn't asking ChatGPT to solve a mystery. I simply wanted help organizing the information I already had.

But that timeline led us somewhere I hadn't expected.

Starting with a Pension File

James Marshall Ricketts served in Company K of the 7th Indiana Cavalry during the Civil War. His pension file contains documents created over many years, making it difficult to see his military story as a whole.

ChatGPT organized those records chronologically.

The resulting timeline showed James enrolling in August 1863, being mustered into service in September, suffering an injury that November, spending time in a Memphis hospital, returning to duty, and eventually being mustered out as a corporal in September 1865.

It also followed his postwar pension claims.

Creating the timeline didn't uncover new records. Instead, it rearranged information from the pension file into a form that made patterns—and inconsistencies—easier to see.

And one inconsistency caught my attention.

Wait—How Old Was James?

The pension records included a birth date of 28 November 1847.

James enlisted on 18 August 1863.

That would make him only 15 years old.

Yet one of his military records reported his age as 18.

At that point, our conversation shifted.

Instead of simply asking what the pension file said about James's military service, I started thinking about what I already knew from census records.

I checked those records and shared the ages with ChatGPT:

• 1850 — age 3

• 1860 — age 12

• 1880 — age 32

• 1885 — age 38

• 1895 — age 48

• 1900 — born November 1848

Those records consistently supported a birth around 1847 or 1848.

Suddenly, my original project wasn't just about creating a military timeline.

It had become an investigation into an underage Civil War soldier.

Bringing Another Record into the Conversation

Fortunately, I also had a copy of James's compiled military service record.

So I uploaded that file and asked another question:

Does this change anything about his Civil War story?

It did.

One of the wartime records—the Company Descriptive Book—actually listed James as 18 years old when he enlisted.

That was important.

Previously, I could have wondered whether the age of 18 appearing in a later military certificate was simply a transcription error or something introduced years after the war. The wartime record showed that wasn't the case.

The Army really did have James recorded as 18.

Yet the census and pension evidence strongly suggested he was only 15.

That changed the question again.

Now it wasn't simply:

How old was James when he enlisted?

It became:

How did a 15-year-old James Marshall Ricketts enter the Union Army as an 18-year-old?

I still don't know the answer.

The records currently available don't tell me whether James personally claimed to be 18, whether a recruiter entered that age, or whether there was some other explanation.

But they gave me a much better research question.

The Military Record Added More Pieces

Looking closely at the military file also strengthened other portions of James's story.

His pension file said that his horse fell on him near Germantown, Tennessee, in November 1863, crushing and bruising his right leg. According to James, he was sent to Overton General Hospital in Memphis.

The military records don't document the horse accident itself.

However, they do document James as sick in a Memphis hospital. Even better, hospital muster rolls place him specifically at Overton U.S.A. General Hospital during the early months of 1864.

Thus, one source supplied the story of the accident while another independently supported the resulting hospitalization.

The military records also showed that James eventually returned to duty.

Then came another interesting discovery.

On 13 December 1864, James was promoted to corporal.

If his November 1847 birth date is correct, he had turned 17 only a couple of weeks earlier.

He continued serving until he was mustered out at Hempstead, Texas, on 19 September 1865.

That meant James's entire Civil War service apparently occurred before his eighteenth birthday.

From Timeline to Story

By this point, my original request for a timeline had evolved considerably.

The conversation had moved through several stages:

Pension transcription → military timeline → age discrepancy → census comparison → military record → underage enlistment → revised Civil War story

And it didn't stop there.

Once we understood James's story better, I asked ChatGPT to help turn the research into a blog post.

Then the story inspired another idea.

Could James's military service become a song?

ChatGPT helped write lyrics for “Fifteen and Riding.” I took those lyrics to Suno and asked it to create a song in the style of a Civil War-era military ballad. I then combined the resulting song with a feature image in Canva and created a video to share on YouTube.

All of that grew from one simple request to organize a pension transcription into a timeline.

ChatGPT as Part of the Research Conversation

This experience illustrates one of the ways I've found ChatGPT most useful in genealogy.

I don't necessarily have to know where the conversation is going when I start.

I can begin with something very specific:

Create a timeline.

Then I can react to what I see.

What about his age?

I can bring additional evidence into the conversation.

Here are the census ages. Does that change the conclusion?

Then another source.

Here's his military record. What does it add?

Each question builds on the previous one.

In that sense, using ChatGPT can resemble having a research conversation. It can help organize evidence, point out discrepancies, compare records, and suggest questions raised by those discrepancies.

But there's an important distinction.

The conversation isn't the evidence.

James's pension file is evidence.

His military service record is evidence.

The census records are evidence.

ChatGPT helped me bring those pieces together and think about what they might mean.

And sometimes that process reveals that the most interesting question isn't the one I started with.

Following Where the Records Lead

When I began this conversation, I wanted a timeline of James Marshall Ricketts's Civil War service.

I ended up looking much more closely at a 15-year-old Indiana farm boy who somehow entered the Union Army with his age recorded as 18, was hospitalized within months, returned to his regiment, became a corporal at about 17, and remained in service until September 1865.

I also ended up with a blog post, a song, a feature image, and a YouTube video.

None of those were part of the original plan.

But perhaps that's what makes both genealogy research and conversations with AI interesting.

One question leads to another. One record changes how we understand another. And sometimes the best discoveries happen when we're willing to follow the conversation—and the evidence—where they lead.