From Indentured Servant to Research Assistant: My Journey with AI

AI Image

When people ask how I use AI in genealogy, they often assume I simply type a question and accept whatever answer appears on the screen.

Nothing could be further from the truth. Our relationship didn’t begin with trust. It began with skepticism.

At first, AI was my indentured servant.

I gave it small, well-defined tasks and watched closely. Summarize this article. Explain this record. Rewrite this paragraph. Find an error. Every response was checked. Every citation was verified. Every conclusion was questioned.

Sometimes it got things wrong. Sometimes it made assumptions. Sometimes it confidently presented information that simply wasn’t supported by the evidence.

As genealogists, we have a name for that. Bad research! So I tested it again.

And again. Eventually, something changed.

I realized that if I clearly explained my expectations, required citations, and insisted that every statement be supported by evidence, the quality improved dramatically. AI wasn’t learning about genealogy in the human sense, but I was learning how to work with it effectively.

That was the apprentice stage.

Instead of assigning isolated tasks, I began involving AI in larger projects. It helped organize research logs, suggest outlines, identify inconsistencies, and point out places where my reasoning could be strengthened. It wasn’t replacing my judgment. It was helping me think more systematically.

Today, I think of AI as a research assistant.

Notice what I didn’t say. I didn’t say expert. I didn’t say genealogist. And I certainly didn’t say author. Those jobs are still mine.

A good research assistant doesn’t decide whether the evidence proves a relationship. It doesn’t determine whether conflicting records can be reconciled. It doesn’t know when a family story should be believed or discarded.

Those decisions require experience, historical context, and professional judgment.

What AI does remarkably well is something different. It organizes information, checks for consistency, synthesizes large amounts of material, and suggests alternative ways of looking at a problem. It catches details I might overlook after staring at the same document for hours.

Most importantly, it gives me someone to argue with.

That may sound odd, but some of my best discoveries have come from asking AI to challenge my assumptions rather than agree with them. If my conclusion is sound, it survives the questioning. If it isn’t, I’d rather discover the weakness before publishing than after.

Looking back, I realize something else changed. In the beginning, I gave AI individual tasks. Today, I delegate entire processes.

Instead of asking it to summarize one article, I ask it to help design a workshop, organize a book, review the consistency of a manuscript, or identify gaps in a research plan. Those are collaborative efforts rather than isolated assignments.

That shift didn’t happen overnight. It was earned through months of testing, refining, correcting, and learning from each other. Ironically, the biggest lesson wasn’t about artificial intelligence at all. It was about partnership.

The best research assistants don’t replace researchers. They make researchers better. Perhaps that’s the real promise of AI for genealogy.

Not replacing the genealogist, but giving each of us another thoughtful set of eyes as we pursue the stories of those who came before us.

Leave a Reply

Your email address will not be published. Required fields are marked *