Real Examples from My Own Research
Recently, Randy Seaver of Genea-Musings asked an artificial intelligence tool an interesting question:
“Please provide a list of examples of how a ‘Thinking AI’ model might be used in June 2026 to help genealogy and family history researchers and writers.”
Randy shared the results in his post, “Ask AI: What Can a Thinking AI Model Do for Genealogy and Family History Researchers?” on Genea-Musings. Randy’s question prompted me to wonder whether I was already using a “thinking AI” model in my own genealogy work.
So I asked ChatGPT a simple question:
“Are you a thinking AI tool?”
The answer was yes. While AI tools are often associated with transcription, translation, and content creation, a thinking AI goes beyond those tasks. Its value lies in analyzing evidence, evaluating competing theories, identifying assumptions, and helping researchers reason through difficult problems.
After reviewing the list of genealogy applications generated by AI, I asked a follow-up question:
“Which of the above tasks have you already done in our various chats?”
The answer surprised me.
More Than Transcription and Writing
Like many genealogists, I first began using AI for tasks such as:
- Transcribing documents
- Summarizing records
- Creating blog posts
- Writing biographies
- Generating historical context
However, over time I found myself using AI in a much different way.
Instead of simply asking AI to create content, I was using it as a research partner.
Evaluating Conflicting Evidence
One of the most common uses has been evaluating conflicting evidence.
Whether working with DAR applications, death certificates, census records, or online trees, AI has helped compare sources and identify inconsistencies.
Rather than simply accepting one record over another, the discussions often focus on:
- Who provided the information?
- When was the record created?
- Which source is likely to be most reliable?
- What additional evidence is needed?
These are the same types of questions genealogists ask when evaluating evidence.
Testing Genealogical Hypotheses
Perhaps the best example is my ongoing research involving the various James Crawford families.
Over the years I have accumulated records for multiple men named James Crawford living in Kentucky and Ohio during the same time period. Determining which records belong to which man is not a simple search problem.
Instead, it requires analyzing:
- Land records
- Tax lists
- Probate records
- Marriage records
- Associates and neighbors
- DNA evidence
AI has helped organize the evidence, compare competing theories, and identify weaknesses in each hypothesis.
It hasn’t solved the problem, but it has helped me think through it.
FAN Club Analysis
Many of my brick-wall projects involve researching Friends, Associates, and Neighbors (the FAN Club principle).
In discussions involving the Crawford, Sellers, Anderson, Douglas, and Maxwell families, AI has helped identify recurring connections and patterns across multiple records.
This type of analysis can be difficult when dealing with dozens of individuals spread across many documents.
DNA Analysis and Interpretation
Another area where AI has been useful is DNA analysis.
While AI cannot replace careful DNA research, it can help evaluate:
- Shared match groups
- Leeds Method clusters
- ThruLines evidence
- Y-DNA results
- Descendant documentation strategies
More importantly, it can help identify where assumptions may be creeping into the analysis.
Detecting Assumptions
Ironically, some of my recent blog posts have focused on AI making assumptions.
For example, Google NotebookLM assumed that Albert Hutchinson was working as a farm laborer simply because he appeared in the household of Ephraim Finch in the 1850 census.
The original report did not say that.
Similarly, a NotebookLM video discussing Angelina Currey Burke displayed imagery associated with Posey County, Indiana while discussing her Kentucky origins.
These experiences led to an important realization:
AI is most valuable when it helps us recognize assumptions—whether those assumptions come from AI, online family trees, published genealogies, or even our own research.
Building Research Plans
Another task I had not fully appreciated was research planning.
When discussing my brick-wall ancestor Hiram M. Currey, AI helped develop a research strategy involving:
- Mexican War records
- Bounty land files
- Probate records
- Court records
- Newspaper research
- California sources
- Guardianship records
Rather than simply suggesting records, it helped prioritize potential sources and organize future research.
Creating Family History Narratives
Of course, AI has also become part of my writing workflow.
Many of my recent projects have involved using AI to help create:
- Ancestor biographies
- Military biographies
- Friday Find posts
- Monday’s Diggings posts
- Historical context sections
- Meta descriptions
- Social media blurbs
These tasks are perhaps the most visible uses of AI, but they are not necessarily the most powerful.
A Research Partner
After reviewing the list, I realized that I am already using AI in ways I had not fully recognized.
The most valuable role is not transcription.
It is not content creation.
It is not even document analysis.
The greatest value comes from having a research partner available at any hour—one that can review evidence, challenge assumptions, suggest alternatives, and help organize complex problems.
AI does not replace genealogical judgment. It cannot determine whether a conclusion is correct.
But it can help us think more carefully about the evidence.
And for genealogists working through difficult problems, that may be one of the most useful tools of all.
For those interested in the original discussion, be sure to read Randy Seaver’s Genea-Musings post asking AI what a “Thinking AI” model can do for genealogy researchers. It is an excellent reminder that the future of AI in genealogy may be less about finding records and more about helping us reason through what those records mean.