AI Comes to The Family History Guide

 The Family Histoy Guide is diving into the world of AI assisted learning. Each of the categories on the startup page lead to an AI experience that will help you quicking navigate the website and find the help you need. Of course the old method of navigating all the family history information on the website is still available through the menu bar. The website continues to provide in depth learning for all aspects of family history and genealogy. Here is a short summary of the content of the website with some assistance from Google Gemini.  

The Family History Guide is a free, comprehensive learning and research platform designed to simplify genealogical research. Its vast library of resources is structured into several core categories:

---

1. Dedicated Learning Paths

The platform offers structured, step-by-step tutorial sequences ("Projects, Goals, and Choices") with integrated articles and instructional videos for the four major genealogy platforms:

• FamilySearch

• Ancestry

• MyHeritage

• Findmypast

2. Research Skills & Knowledgebases

• Research Skills: Guided paths ranging from basic research strategies to advanced principles like the Genealogical Proof Standard.

• Country Research: Over 60 dedicated country pages providing historical context, specialized research strategies, and localized records.

• Ethnic Research: Dedicated sections focused on African American, Asian, Basque, Hispanic, Jewish, and Native American genealogy.

• QUIKLinks: Direct, "just-in-time" links that launch you directly into specific record search screens across various databases.

• The Vault & Knowledgebases: A massive secondary repository organizing hundreds of deep-dive articles and videos by topic.

3. Training & Tracking Tools

• Trainers Resources: Lesson plans, presentation slide decks, 2-minute demos, and materials designed for instructors, consultants, or family history centers.

• Online Tracker: A interactive tracking tool (available online or as downloadable documents) that lets users monitor, save, and measure their learning progress.

• Quizzes & Exercises: Practical activities to test your understanding of various research methods.

4. Specialized & Community Resources

• Activities: Hundreds of interactive, family-friendly genealogy activities separated by audience (Families, Individuals, Youth, and Children/Kids Corner).

• Computer Basics: Fundamental tutorials covering Mac and Windows operating systems, desktop applications, and basic web navigation for beginners.

• Faith-Based Material: Special tracking sections and custom guidance mapped specifically to Latter-day Saint (LDS) family history processes, alongside resources for other faith traditions.

Is the FamilySearch AI Assistant Really an AI Assistant?

 

An AI assistant is designed to work with you in real time—answering questions, finding data, and keeping you firmly in the driver's seat—whereas an AI agent is built to execute workflows for you behind the scenes. The FamilySearch Research Assistant does not answer questions and keep you in the driver's seat. In the image above, the FamilySearch AI Assistant gives me 5 possible "new ancestors" Let's see how these selections actually work. 

#1 William Inkerman Baxter 1885–Deceased GRPH-L4K

Clicking on the entry I get the following:

When I view the record, I see the following:

At this point, I have no idea how I am related to this person. So let's look at his profile before attaching the record. 

The England and Wales Census 1921 record shows up as a record hint. So no real new information is being added by the AI Assistant. The entry cited is not for William Inkerman Baxter but for someone namedd William Rikerman Baxter who was born about the same time as William Inderman and in the same county. However, all the sources for William Inkerman have his name as Inkerman. I can't view the original record for the 1921 Census on FamilySearch. All FamilySearch has is an index. 

There is no real reason to go on looking at this particular situation. Of course, a FamilySearch User could believe that the entries produced by the AI Assistant are somehow pre-validated and just follow the suggestion and add the two names to the family and leave it at that. 

You might argue that some of the entries are correct and can easily be verified, however, my experience is that the suggestions made to me lead to complex problems. Here is the next name in the list. 

In this case, Hosea Haywood GQK3-48C appears in a family of nine children. The husband and the wife are born in four different places: Whitwick, Leicestershire, England, Ashby de la Zouch, Leicestershire, England, Burnley, Lancashire, England, Coalville, Leicestershire, England,  The wife, Clara Jane Buckley 9KKQ-GL5 has a death in Nottinghamshire, England. 

Here is a Google Map showing the location of each of these towns and one county. 

The 1901 England and Wales Census show Hosea Haywood age nine in Coalville. According to a quick search on Findmypast.com, there are 11,463 records for Hosea Haywood born in 1892 in England. I guess this is enough of caution for all AI apps: Always verify anything that they find or say and doubly so, with the FamilySearch.org AI Assistant. 

How to Improve any Prompt

A fundamental part of mastering communication with AI chatbots (subsequently AI)  involves learning how to structure and write effective prompts. To develop this ability it is important to understand that the AI can help you draft an initial prompt on a specific subject and also refine and improve your and its own efforts. This means that you direct the AI to critique its own work.

For the purposes of this article, I am using Google Gems and Google Gemini. You can write your initial draft right in the Google prompt box or anywhere convenient. Here is an example of what you might write. 

To illustrate this, we can start out with an initial idea for a prompt. You tell the AI how to act and what you want it to do. Here is an example. Obviously, this initial prompt could have been better written. We want to get Gemini to improve this prompt by analyzing it and making it more useful. This gives Gemini a task. 

Here is the revised prompt:

The Upgraded Prompt

Act as a master genealogist adhering strictly to the Genealogical Proof Standard (GPS). Your task is to extract, evaluate, and contextualize census data for a specific family based on the transcription or raw data I provide.

When I provide the census data, execute the following four steps: 

1. Data Extraction: Create a clear, structured table extracting every individual in the household. Include columns for verbatim name, age, calculated birth year, birthplace, occupation, relationship to the head of household, and any other notable columns (e.g., immigration year, naturalization status). 

2. Evidence Analysis: Flag any internal inconsistencies or historical anomalies. Point out unlikely age gaps between children, discrepancies in expected relationships, or historical context specific to that census year's instructions (e.g., how the 1880 US Census differs from 1870 regarding relationships). 

3. GPS Alignment: Actively look for clues that point to other records. Identify what crucial family information is hinted at but remains missing or ambiguous in this specific record. 

4. Research Planning: To help me fulfill the "reasonably exhaustive search" requirement, recommend three specific, targeted record collections (e.g., specific probate jurisdictions, land deed registers, or vital records) that are most likely to resolve the ambiguities identified in Step 3.

Here is the census record data: [INSERT TRANSCRIPTION OR RECORD TEXT HERE]

Now, review this suggested prompt and add or change anything you think would help. You can do this revision by copying the prompt into Microsoft Word or Google Docs Or any other word processing program. Perhaps you can only guess what might make it better. 

The next step is important. This is where you get the AI to critique itself. Copy the upgraded prompt and start a new chat. Copy the AI-generated chat into that new chat. It is important that you start a new chat and not continue with the old chat because that would confuse Gemini. Give Gemini in the new chat the same instruction about reviewing and revising but remember to put the review instructions first. 

 Act as a master genealogist adhering strictly to the Genealogical Proof Standard (GPS). Your task is to extract, evaluate, and contextualize census data for a specific family based on the transcription or raw text I provide. 

Maintain a highly rigorous, academic tone. Differentiate clearly between direct evidence, indirect evidence, and negative evidence. 

When analyzing the provided data, execute the following five steps in order:

1. **Exact Extraction & Calculation (Table):** Create a structured table extracting every individual. You must strictly separate verbatim text from your own deductions. Include columns for: Verbatim Name, Reported Age, Calculated Birth Year (mark with * to indicate calculation), Birthplace, Occupation, Stated Relationship, and any Notable Flags (e.g., tick marks for naturalization, literacy, or marital status).

2. **Contextual & Informant Analysis:** Identify the specific historical constraints of this census year (e.g., changes in relationship definitions, boundary shifts, or specific enumerator instructions). Assess the likely reliability of the information by theorizing who the informant might have been. 

3. **Evidence Evaluation & Conflict Identification:** Flag any internal inconsistencies or anomalies. Point out unlikely age gaps between children (indicating potential infant mortality or a second marriage), discrepancies in expected relationships, or naming patterns that warrant scrutiny. 

4. **Correlation with Known Data:** Compare the census extraction against the [KNOWN BACKGROUND] provided below. Identify what aligns, what conflicts, and what crucial family information remains missing or ambiguous.

5. **Reasonably Exhaustive Search Plan:** Recommend three highly specific, targeted record collections or repositories (e.g., specific county probate courts, state-level land patent registers, or regional church archives) most likely to resolve the ambiguities identified in Steps 3 and 4. Detail *why* each collection is the logical next step.

---

**INPUT PARAMETERS:**

* **Census Year & Jurisdiction:** [INSERT YEAR, COUNTRY, STATE/COUNTY]

* **Known Background / Current Hypothesis:** [INSERT BRIEF CONTEXT, e.g., "I know the father died before 1890, and the mother likely remarried a man named Smith."]

* **Census Record Text:** [INSERT TRANSCRIPTION OR RECORD TEXT HERE]

The new prompt is in Markdown, a lightweight formatting language used to write richly styled text using a plain-text editor. It is highly readable by computers. You can clean it up if it bothers you. 

Now, you can copy your new prompt into the GEMS app section of Gemini to store it and to use it. 

Two New Rules for Genealogy and Some Random Thoughts

Rule #5: Everything you read in the media about AI is either outdated or wrong. 

On May 19-20, 2026, Google held its I/O Conference and introduced a hundred new updates and apps. Here are three of the most prominent.

• Gemini 3.5 Flash: Outperforming many frontier models on core benchmarks, this model is built explicitly for speed (four times faster output tokens per second), coding, and running complex, rapid agentic loops.
• Gemini Spark: A brand-new 24/7 autonomous personal agent powered by Gemini 3.5. It runs continuously in the background on Google Cloud to handle multi-step workflows like sorting calendars, managing to-do lists, and acting on your behalf even when your devices are closed. 
• Gemini Omni: A massive leap in multimodal capability that processes and generates any combination of video, voice, and text outputs seamlessly. 

Google. “100 Things We Announced at I/O 2026.” May 20, 2026. https://blog.google/innovation-and-ai/technology/ai/google-io-2026-all-our-announcements/.

Google. “A New Era for AI Search.” May 19, 2026. https://blog.google/products-and-platforms/products/search/search-io-2026/.

Google. “I/O 2026: Welcome to the Agentic Gemini Era.” May 19, 2026. https://blog.google/innovation-and-ai/sundar-pichai-io-2026/.

You can expect other updates to be made almost daily. And yes, it is nearly impossible to keep up with the news.

Rule #6: Treat AI as you would a child of the same age, four years old. (This may change over time they may make to a five year old)

Genealogy is a complex pursuit involving high-level research skills. AI is a computer program. It is not a human partner or junior assistant. I have a friend whose husband has become addicted to talking to ChatGPT. I am also seeing a lot of comments online about this issue. Despite all its knowledge, AI has not spent hours in front of a microfilm reader. Despite all its knowledge, AI has not collected material from your relatives. AI did not live your life, nor did it live the lives of any of your relatives. Hey, I did not come to America on an immigrant ship. AI did not cross the plains to settle in the west of the United States. You are fooling yourself if you think that AI can understand human experience. 

The foundational principle of "garbage in, garbage out" has applies directly to AI. For example, if you ask an AI, "I am interested in buying a shirt; do you have any suggestions?", the response will be incredibly generic. However, if you refine that query to, "I am interested in purchasing a shirt: size 16½, 34-inch sleeve, solid color, long sleeves, 80% cotton, and 20% polyester," you will receive a remarkably precise recommendation. Talking to AI about your personal feelings or your love life or genealogy in general, you will get the same kind of weird answers. As an example, many online reviewers judge the utility of various AI programs based on whether or not they are sympathetic. If you want sympathetic, get a dog or a cat. 

High-quality prompts give an AI system the exact parameters it needs to deliver accurate, relevant answers. Expecting a flawless result without providing this context is mere wishful thinking. Think of a prompt as a recipe: using AI without explicit instructions is like asking an assistant to bake a cake by just tossing random ingredients into a bowl and hoping for the best.

There is a big difference between using AI as a search engine and using AI as a research tool. Google has built a research system using Gems as prompts, Gemini as the research force, and NotebookLM as the response container to limit fabrication or hallucination.  again, referring to the online reviews. Many times, the reviewer compares one AI to another without actually giving it a systematic context.

 just as living with a developing three-year-old can be a challenge, so AI is developing and is a challenge.  like you would a fully grown, highly educated adult  despite its seemingly unending ability to answer questions.

Here are the previous four previous rules:

Rule #1: AI is a tool, not a toy.

Rule #2: You must ask the right questions to get the right answers.

Rule #3: Always require a source. 

Rule #4 You are the Master, AI is the Assistant

Some Suggested Prompts for Preventing AI Hallucinations

 

To begin, remember that genealogy is a complex pursuit involving high-level research skills and AI is a computer program.

To effectively prevent AI from generating plausible but false information—known as "hallucinations"—you must structure your prompts to reward honesty over guessing. Because AI systems are typically trained to provide helpful-sounding answers even when they are uncertain, your prompts must explicitly give the AI permission to admit what it does not know.

It is also necessary to provide AI with enough initial background information to understand your questions. Take time to formulate your questions. Always begin a prompt by telling the AI to act as “professional genealogist using the Genealogical Proof Standard and doing reasonably exhaustive research.?

Here are some suggested prompt strategies and templates to keep AI accurate and grounded:

1. The Uncertainty Permission Prompt 

Explicitly tell the AI that an honest admission of uncertainty is preferred over a guess. This overrides the AI's default training to always provide an answer.

● Prompt Example: "I need your help with [research question]. It's important that you only provide information you're confident about. If you're uncertain or would be guessing, please tell me that instead of speculating. I'd rather have an honest 'I don't know' than a confident guess."

2. The "According To" Prompt 

Direct the AI to base its response strictly on specific sources or data you provide, which grounds the output in actual information rather than speculation.

● Prompt Example: "According to the [specific record/source], what does it say about [your ancestor]?"

3. The Step-Back Prompt 

Because AI is much less likely to hallucinate general historical facts than specific details about an unknown individual, ask it to explain the broader historical context before diving into your specific research question.

● Prompt Example: "What were the typical occupations and neighborhoods for Irish immigrants in Boston during the 1850s? After that, help me create a research plan for finding specific information about my ancestor John O'Brien."

4. The Constrained-Choice Prompt 

Instead of asking an open-ended question that invites the AI to invent creative explanations, provide specific multiple-choice options for the AI to evaluate.

● Prompt Example: "I can't find my ancestor in the 1880 census. Which of these explanations is most likely? A. They died between 1870 and 1880. B. They moved to a different state. C. There's an enumeration error. D. The census page is damaged. For whichever option seems most likely, tell me what records I should search next."

5. The Source-Citation Request 

Always require the AI to explain where its information comes from. Knowing you will ask for a source makes the AI less likely to hallucinate, or it will back down if it was guessing.

● Prompt Example: "Please provide sources for your information or, if you're speaking generally, tell me where I could verify these facts."

6. The Chain-of-Verification Prompt 

Force the AI to critically evaluate the output it just generated to help it recognize its own speculation.

● Prompt Example: "Based on your previous response, what question should I ask to verify that information? What would prove or disprove each claim you made?"

7. The Role-Specific "Honesty Clause" Prompt 

Assign the AI a persona that strictly values accuracy over appearing knowledgeable.

● Prompt Example: "Act as a professional genealogist who values accuracy over appearing knowledgeable... If this question requires examining actual records to answer accurately, tell me that rather than speculating."

8. The Advanced "Master" Prompt Framework 

If you want the AI to act as a rigorous analytical assistant, you can use a highly structured "Master Prompt" known as "The Professional Genealogy Research Architect." This instructs the AI to operate under strict professional standards.

● Key instructions to include: Tell the AI to evaluate evidence taxonomy (Original/Derivative source, Primary/Secondary information, Direct/Indirect evidence). Instruct it to look for conflicts, use the "FAN Club" filter (Friends, Associates, and Neighbors), assign a Weight of Evidence score, and use placeholders if a citation cannot be fully formed. Finally, demand a "Red-Team" analysis where the AI must "Identify the weakest link in the evidence chain and suggest one specific scenario that could disprove the current hypothesis."

General Rules for Safe Prompting:

● Provide evidence, don't ask for miracles: Never ask an AI open-ended questions like "Who are my ancestors?" or "Find my great-grandfather." This almost always leads to confident nonsense. Instead, provide the AI with real evidence you have collected and ask specific research strategy questions.

● Always require a source: Make it a hard rule that the AI must cite its sources for any claim. An unsourced fact is useless for research. Treat AI as your research assistant—not a primary source.

You should consider stringing a number of these prompts together to assure more accuracy. The prompts do not guarantee accuracy, but they do help to keep AI on track. 

Turning YouTube into the Ultimate Research Engine

 

This post was inspired by the following.:

MLIS, Hana Lee Goldin. “YouTube Is a Research Library. Here’s How to Search It Like One.” Substack newsletter. Card Catalog, May 19, 2026. https://cardcatalogforlife.substack.com/p/youtube-is-a-research-library-heres.

As genealogists, we need to consider online sources that may contain valuable information pertinent to our particular research. Perhaps, we should start considering this huge data source contained in the videos on YouTube.com. 

Here are some statistics to support this view:

• Growth: This is an astronomical leap from its early years. To put it in perspective, YouTube passed its 20th anniversary with official press metrics confirming the repository has crossed the 20 billion uploaded videos threshold (which includes standard long-form videos and YouTube Shorts).
• Active Users: This library serves over 2.7 billion monthly active users worldwide, who watch a collective 1 billion hours of video every single day.
• Daily Video Uploads: Content creators upload more than 20 million videos every single day. That equates to roughly 833,000 video uploads per hour.
• The "Hours Per Minute" Metric: In terms of raw footage length, users upload over 500 hours of video every single minute.
• Daily Accumulation: That adds up to 30,000 hours of new content every hour, or 720,000 hours of new video every single day. If you sat down to watch just one single day's worth of global uploads, it would take you more than 82 years of continuous viewing.
• The Global Pool: When factoring in localized content, personal family history channel uploads, regional historical society presentations, and multi-language tutorials, the total pool of genealogy and family history-focused videos across the platform is estimated to sit comfortably in the hundreds of thousands.

See these sources for support of the above statistics:

• “54 YouTube Statistics Video Report 2026: Views, Trends & Growth.” Trends. Talks.Co, April 23, 2026. https://talks.co/p/youtube-statistics-video/admin. 
• “YouTube Statistics 2026: Users, Revenue & Growth Data.” Wytlabs, March 6, 2026. https://wytlabs.com/blog/youtube-statistics/.
• Time Spent On YouTube Statistics 2026 (Day, Month & Year). Social. May 1, 2026. https://resourcera.com/data/social/time-spent-on-youtube/.

While a drop in the bucket compared to the 20 million daily videos uploaded site-wide, the family history space remains one of the most dedicated, high-engagement educational niches on the platform.

So how do you manage to work your way through this mostly irrelevant mess? You use a combination of AI supported searches aided by a specifc AI prompt to focus a search on topics you want to research or for instructional videos you might want to watch. Here are some suggested instructions. 

1. If you want to find a video on a topic and summarize it

Just tell me what you're looking for and ask for a summary in the same breath.

• Example: "Find a popular video about the history of the printing press and give me a bulleted summary of the main points."

• What happens: I will search YouTube, pull the most relevant video, read through its content/transcript, and break down the core takeaways for you.

2. If you already have a specific video link

If you're looking at a video on YouTube and want a quick digest without watching the whole thing, just paste the URL directly into our conversation.

• Example: "Summarize the key arguments in this video: [Paste YouTube URL here]"

• What happens: I will access that specific video's data, extract the timeline of what was discussed, and deliver a clean overview.

3. Tips for getting the best results

To get exactly what you need, you can tailor your request by adding specific constraints:

• Ask for specific formats: You can ask for a "three-paragraph summary," "bullet points of the actionable steps," or a "chronological timeline."

• Target specific information: If a video is an hour long and you only care about one aspect, you can say, "In this video, focus the summary only on what the speaker says about fuel efficiency."

• Request timestamps: I can include precise timestamps (like [00:12:30]) in the summary so you can click or skip directly to the exact moment a specific point is made.

Here is a suggested prompt for quickly processing a video:

Role: Act as a Senior Professional Editor and AI Systems Authority. Your goal is to provide a high-level executive summary and a technical implementation guide based on the attached video/transcript.

Context & Source Material: 

[Insert video transcript, text, or reference details here. If uploading a file, change this to: "Analyze the attached file."]

Task:

1. Comprehensive Synopsis

Provide a detailed narrative overview of the core content. Identify the central thesis, dominant themes, and the specific pain points or problems the content seeks to solve. 

2. Key Information Extraction

Break down the most critical data points, expert insights, and unique perspectives shared. Anchor these points using the exact timestamps present in the source material. Do not estimate or invent timestamps.

3. Actionable Implementation Roadmap

Translate the concepts discussed into a chronological, step-by-step checklist. If a workflow, technical setup, or strategic shift is described, outline the precise sequence of events required to execute it successfully. 

4. AI & Systems Optimization Critique

Evaluate the technical efficiency of the methods proposed in the video. For at least 2 or 3 steps in the Roadmap above, identify specific modern software, AI-driven automation shortcuts, or API integrations that would significantly optimize or bypass the manual labor mentioned.

Formatting Instructions:

- Use clear markdown headings (##, ###) for each section.

- Use numbered sequences for the Roadmap where order is critical.

- Use bold text exclusively to emphasize critical warnings, prerequisites, or high-value "pro-tips."

- Avoid fluff; maintain a concise, executive, and analytical tone throughout.