Skip to content

Using AI in Research: Always Verify Your Sources!

  • by Mark Rabideau- Many-Roads Genealogy https://many-roads.com
  • eirenicon llc.
  • 711 Nob Hill Trail - Franktown, Colorado 80116
  • +1.303.660.9400
  • genealogy@many-roads.com

A Crucial Note for All AI Users in Genealogy & History

Artificial Intelligence (AI) tools are incredibly powerful and can significantly accelerate your genealogical and historical research—like a turbo boost for digging through dusty records. They can help with transcription, summarization, language translation, and even suggest connections (e.g., "This census entry might link to your Irish roots"). However, it's vital to approach AI-generated information with a critical eye, just like you'd double-check a family story told at Thanksgiving.

Why is Verification So Critical?

  1. AI "Hallucinations": General-purpose AI models, especially large language models (LLMs), can sometimes generate information that sounds perfectly plausible but is entirely incorrect or made up. This phenomenon is often called "hallucination" (see Handout #1). They predict what should come next based on their training data, not always what is factually true—e.g., inventing a "lost diary" for your great-aunt or claiming a 1920s ancestor used email.
  2. Outdated or Biased Data: AI models are trained on vast datasets, with some updated to mid-2025, but they may still miss the latest findings or reflect biases from older sources. For instance, they might perpetuate outdated migration patterns unless cross-checked with 2025 genealogical updates.
  3. Lack of Understanding: AI doesn't "understand" history or family relationships in the human sense. It processes patterns and probabilities, like a super-smart parrot repeating phrases. It cannot discern truth, reliability, or nuance like a human researcher can—missing the "why" behind a name change.
  4. No Inherent Citing: While some AI tools (e.g., Perplexity, Claude) provide sources in 2025, many general-purpose ones do not, or they offer generic or invented citations (e.g., a fake book title that doesn't exist).

Your Role: The Ultimate Arbiter of Truth!

As the human researcher, you remain the ultimate authority and quality controller of your findings. You're the family historian with the wisdom of years—AI is just the eager helper carrying the books.

  • Cross-Reference Relentlessly: Always verify any information provided by AI with multiple independent, reliable sources. Consult original documents (birth certificates, photos), reputable archival records, academic publications, and established genealogical databases like FamilySearch or MyHeritage.
  • Prioritize Original Sources: Use AI as a starting point or an assistant (from Handout #2), but always seek out and examine the primary and secondary sources it references—or that you would typically consult. Tools like FamilySearch's Full-Text Search AI (https://www.familysearch.org/en/labs/full-text-search) make this easier by linking to unindexed records.
  • Critical Thinking First: Do not accept AI output at face value. Apply your historical knowledge, logical reasoning, and genealogical best practices to evaluate its suggestions—ask: "Does this fit what I already know? Where's the proof?" For 2025, test AI claims against FamilySearch’s 2025 AI-enhanced record updates.

Think of AI as a skilled research assistant who needs constant supervision and fact-checking. It can help you find leads and organize information, but the responsibility for accuracy and truth always rests with you. Build your tree on facts, not guesses—and you'll have stories that last generations!

  • Verification Guides:
  • FamilySearch's "Verify AI in Genealogy" Checklist: https://www.familysearch.org/en/blog/free-familysearch-webinars-august-2025 (Includes 2025 AI verification sessions with live demos—free printable.)

  • ICAPGen's Best Practices: https://www.icapgen.org/resources/ai-in-genealogy-verification/ (From certified genealogists; updated tips on cross-referencing for 2025.)

  • Bias & Limitations Explainers:

  • Harvard Library's AI for Historians: https://guides.library.harvard.edu/ai-for-historians (Updated 2025 case studies on biases, including genealogy-specific examples.)

  • BBC Bitesize on AI Hallucinations: https://www.youtube.com/watch?v=rYWEl6OSIBE (Getting Started with AI in 2025—beginner-friendly video under 3 minutes.)

  • Quick Tools for Checking:

  • Snopes for Historical Myths: https://www.snopes.com/fact-check/ (Updated 2025 database to test AI folklore claims.)
  • Google Books Ngram Viewer: https://books.google.com/ngrams (Free to spot if a phrase/event is anachronistic, with 2025 data integration.)

Credits

Compiled by Mark Rabideau, Opa & Professional Genealogist.