AI assistants respond to the specificity and clarity of your questions; vague prompts yield generic answers, while precise ones that name dates, places, and what you've already ruled out produce more useful research suggestions. Learning to frame genealogical questions as concrete problems rather than open-ended inquiries makes AI tools genuinely productive.
There's a big difference between asking an AI "Tell me about my ancestor" and crafting a question that actually gets you useful genealogical information. This skill is called prompt engineering—basically, it's the art of asking AI questions in ways that lead to better answers.
Think of it like the difference between asking a librarian "Do you have books?" versus "I'm looking for passenger lists from ships arriving in Boston in 1912." The second question is more likely to get you exactly what you need.
In genealogy, prompt engineering matters because AI doesn't automatically know the context of your family history, the time period you're researching, or what kinds of documents might exist. You have to give it that context. Here's what makes a genealogy prompt work:
Include specific details: Instead of "What can you tell me about my ancestor John Smith?" try "My great-grandfather John Smith immigrated to Boston in 1903 from County Cork, Ireland. He worked as a laborer. What documents should I look for to verify his immigration date and find his siblings?"
Name the document type: AI thinks differently when you specify you're looking for census records, naturalization papers, or church registries. Say "What would a 1920 census record tell me about John Smith's household composition?" rather than asking vaguely about his life.
Ask for multiple research angles: Instead of a yes-or-no question, ask: "What are three different ways I could verify this ancestor's birth date in 1845, and where would I find that information?" This forces the AI to think like a genealogist and give you actionable leads.
Highlight fragmentary information: If you only know partial facts (like "lived in Pennsylvania" or "had a child born 1887"), explicitly frame that for the AI: "I know my ancestor lived in Pennsylvania and had a daughter named Catherine born in 1887, but nothing else. What research strategy would help me find her parents' names?"
The underlying reason this works: AI is designed to respond to specificity. The more context you provide, the more the AI understands your actual genealogical problem, not just the bare words you typed. When you craft a prompt like a genealogist would think about a case, the AI responds like a genealogist too.
Common mistake: asking open-ended questions expecting the AI to know what genealogical problem you're solving. The AI doesn't know you're stuck. It doesn't know which sources you've already checked or what era your family emigrated. You have to paint that picture.
Try this: Take one ancestor you're researching and write two versions of a question about them. First version: a simple, generic question. Second version: a detailed prompt that includes what you know, what you're trying to verify, and what document type you're interested in. Ask both to ChatGPT or Claude and compare which answer is actually useful for your genealogy work.
Peri can explain this concept, give practical examples, help you decide whether it applies to your situation, or recommend a journey if appropriate.
Explore related journeys or tell Peri what you're working through.