Developing an ethical and safe use of AI for genealogy.

 https://youtu.be/R0bfWAYx-OE?si=IvAVbsJAo4zSV6Zd

It is a remarkable time to be a genealogist. For those of us who have spent decades cranking through microfilm, squinting at faded ink, and meticulously tracking down obscure court records, the current shift in the genealogical landscape feels both exhilarating and, for some, a bit unsettling. We are moving out of an era where our primary challenge was simply finding the data and into an era where the challenge is managing the sheer volume of information that modern tools can now uncover. I’ve been thinking a lot lately about how we bridge the gap between the traditional standards we hold dear—the accuracy, the citations, the "old school" rigor—and these "new school" tools like Artificial Intelligence that are fundamentally changing how we work.

The reality we have to face is that the role of the researcher is evolving. We are shifting from being mere data gatherers to becoming forensic auditors. In the past, we spent eighty percent of our time just trying to locate a record. Today, with tools like FamilySearch’s Full Text Search or AI-powered handwriting recognition, the records are finding us. Our job now is to look at the output of these machines with a critical, expert eye to ensure that what is being presented is historically accurate and meets the Genealogical Proof Standard.

One of the most frequent concerns I hear from fellow researchers is a fear of "AI hallucinations" or fabrications. It is a valid concern. If you treat a chatbot like a simple search engine—asking it "who was my great-grandfather"—you are likely to get a mix of facts and creative fiction. But the secret to using these tools effectively lies in understanding that AI is a tool, not a toy, and certainly not a replacement for your own judgment. The key is learning how to talk to it. In the tech world, they call this "prompting," but for us, it’s really about giving specific, professional instructions.

If you want to start using these tools responsibly, I suggest beginning with Google Gemini or a specific application called NotebookLM. The reason I point toward these is simple: memory and control. When you use a program like NotebookLM, you are essentially creating a "closed loop." You upload your own verified documents—your census records, your deed transcriptions, your family histories—and tell the AI to only answer questions based on those specific sources. This effectively eliminates the risk of the AI "making things up" from the vast reaches of the internet. You become the supervisor, and the AI becomes your highly efficient research assistant.

For those ready to take an actionable step today, I encourage you to look at the "Full Text" search option under the search menu on FamilySearch. This tool uses AI to search every word in every document—not just the indexed names. It can find your ancestors buried in the middle of a land deed or a cattle brand registration that would have been impossible to find a few years ago without reading every page of a digital film. When you find a transcription made by AI, treat it as a "Grade 3" or "Probable Draft." It’s a wonderful head start that might do 98% of the work for you, but you still owe it to your ancestors to check that final 2% against the original image.

We must also be careful with how we handle images. With AI, it’s now possible to "generate" a photo of what we think a 1790s farmhouse looked like. While these are great for illustrating a blog post or a family story, we must be ethically clear: always label them as AI-generated illustrations. They are not evidence. We must maintain the integrity of our historical record, ensuring that future generations can distinguish between a verified historical photograph and a modern digital recreation.

As we move forward, don’t be afraid of the technology. I remember when people were afraid to move from paper to computers, or from microfilm to digital images. Each shift felt like a loss of "the old way," but in reality, it just gave us better eyes. AI is simply the next set of eyes. If we direct it with the same precision and ethics we’ve used for a century, we will find that the stories of our families are more accessible than we ever dreamed possible. It’s not about changing why we do genealogy; it’s just about upgrading the tools we use to do it right.

---

Summary of Resources & Next Steps

I've outlined a few ways to get started with these tools in a responsible manner. If you're looking to dive deeper into the technical side of what I've discussed, here are the primary avenues I recommend:

• FamilySearch Labs: Explore the "Full Text Search" to find names within the body of unindexed records.

• Google Gemini: Practice "prompting" by telling the AI to "Act as a professional genealogist" before asking it to analyze a transcript.

• NotebookLM: Use this to upload your own PDFs and research notes to create a controlled environment for AI analysis.

• CRAIG (Coalition for Responsible AI in Genealogy): Visit cragon.org to read the mission statement and guidelines for ethical AI use.

---

Note: The insights in this post were adapted from the BYU Library Family History Center presentation by James Tanner on the ethical and safe use of AI for genealogy.