What Does AI Know About Me?

 Other genealogists have been asking ChatGPT what AI knows about them, so I asked.

If you asked me, “What does AI know about me?”, based on our conversations, I could actually give a fairly detailed answer—not because I have searched the internet for you, but because of what you have shared while we have worked together.

Here is how I would describe you:

What I know about you

You are Marcia Philbrick, a retired educator, librarian, and technology coordinator in northeast Kansas. Technology has been part of your professional life for decades, so your interest in AI is really a continuation of something you've been doing for a long time: figuring out how new technology can be put to practical use.

Genealogy is one of your major interests. You maintain Heartland Genealogy, where you write about your research, discoveries, records, DNA evidence, research problems, and the process of doing genealogy. You don't just want names and dates—you are interested in reconstructing people's lives and understanding the communities, geography, history, and relationships surrounding them.

I know quite a bit about your Crawford research in particular. We've discussed multiple James Crawfords in Kentucky, Virginia, Indiana, and Ohio; the Crawford–Sellers connections; land records; marriage bonds; FAN networks; and the complications created by Y-DNA results. I also know about some of your persistent research problems, including the parents of Sally/Sarah Smith, the identity and relationships of various Crawford families, and the disappearance of Hiram M. Currey.

More recently, I've followed your work on Albert Hutchinson. You've used traditional records, Ancestry DNA matches, shared-match networks, ThruLines, and Y-DNA testing to reconsider his parentage. Importantly, I've learned something about your research philosophy from that work: you're willing to undo an accepted family relationship when the evidence no longer supports it rather than trying to make new evidence fit an old conclusion.

I also know quite a bit about how you work. You use RootsMagic, Ancestry, FamilySearch, WikiTree, GEDmatch and other genealogy resources. You frequently bring records to me for transcription, analysis, comparison, or help turning the evidence into a narrative. You've also used AI for brainstorming research strategies, analyzing networks, understanding DNA, interpreting historical documents, and creating blog posts and graphics.

Your AI "prompting style" is interesting in itself. Although you occasionally use a detailed prompt—particularly for biographies—you generally prefer the KISS approach. Many productive conversations begin with something as simple as “transcribe this,” “can you improve this?” or a question about something you've encountered in your research. You tend to use AI conversationally, building on the answer rather than trying to construct one enormous perfect prompt.

Outside genealogy, I know that you volunteer with the Nemaha County Historical Society, where you've worked on exhibits, museum signage, fundraising publicity, educational materials, preservation projects, and local-history research. You've also asked me to help explain artifacts such as wool carders and spinning wheels for museum visitors.

I know that you are involved with Seneca United Methodist Church, including preparing worship presentation materials, and that your husband, Mike, has helped with recording services.

And I know something about your family because family history naturally overlaps with family life. You've mentioned your husband Mike, your children and grandchildren, and occasionally your own memories—such as your experience as a school librarian/technology coordinator on September 11, 2001.

But there is an important distinction

There are really three different meanings hiding inside the question “What does AI know about me?”

What ChatGPT has learned from our conversations is what I've described above. It is surprisingly extensive because we've had many conversations and you repeatedly work with me on related projects.

What an AI could discover about you from the public internet is a different question. That might include your blog, genealogy profiles, historical-society activities, presentations, social-media material, and other publicly indexed information.

And what the underlying AI model knew about you before our conversations is different again. I don't have a way to inspect the model's training data and say, “Here are the things it was trained on about Marcia Philbrick.”

That distinction is important when genealogists conduct the “What does AI know about me?” experiment. An answer from a chatbot can be a mixture of remembered conversational context, information retrieved from the web if searching is used, and general model knowledge. It can also contain inferences—and sometimes errors.

What I think I've learned about you

Perhaps the most interesting part isn't the collection of facts.

From our work together, I would characterize you as a researcher first and a collector of ancestors second. You are uncomfortable declaring a genealogical relationship proven simply because it appears in numerous online trees. You repeatedly return to the underlying records and ask, essentially, “What does the evidence actually say?”

You also seem particularly interested in the research journey. A court record that creates another question isn't necessarily a failure. It can become a blog post about the puzzle. DNA that breaks an established family connection isn't merely frustrating; it becomes another avenue of investigation.

And increasingly, AI has become another tool in that process—not something that does the genealogy for you, but something you converse with while doing genealogy.

That may actually be the most revealing answer to “What does AI know about me?”

After enough conversations, AI doesn't merely accumulate facts about a person. It can begin to recognize how that person approaches a problem, what evidence matters to them, what questions they repeatedly pursue, and how they prefer to communicate their discoveries.

And that raises a much more interesting question for genealogists:

How much of what AI “knows” about us did it find—and how much did we teach it through our conversations?