Introduction to NotebookLM
Today I tried out Google’s “experimental” tool called NotebookLM. I learned about it from a YouTube video: “Google NotebookLM Just Changed Note-taking Forever.”
Google describes NotebookLM, powered by Google’s Gemini AI, as an experimental product designed to leverage the power of language models along with your existing content to gain critical insights faster. It acts like a virtual research assistant that can summarize facts, explain complex ideas, and brainstorm new connections—all based on the sources you select.
The ability to enter sources and receive summaries, ask questions, and forge new connections—as described by Google—highlights one of the key differences with other generative AIs. Since we are providing all the information, this should help mitigate the creation of inaccuracies.
Testing NotebookLM with a Real Genealogy Project
I decided to test it with a genealogy project I am working on for a friend, researching the life of Felix Buster Tiger, who was Creek. I added eight sources to the “Tiger Family” notebook:
- 1910 Census
- 1920 Census
- 1945 newspaper clipping about Pvt. Felix Tiger being released from captivity during WWII
- 1966 newspaper clipping about the descendants of George Tiger, including Felix Buster, and their rights to an oil & gas lease
- 1997 obituary of Felix Buster Tiger
- Dawes Creek card for Felix Buster’s mother’s family (Katie Tiger)
- Dawes Creek card for Felix Buster’s father’s family (George Tiger)
- Proof of heirship for Katie Thompson, Felix Buster’s mother, who had remarried
Transcribing Documents with Claude 3 Opus
I used Claude 3 Opus to transcribe the newspaper clippings; it did an incredible job! I typed the Dawes cards and proof of heirship records. I copied and pasted the summaries of the census records from Ancestry.
Asking Questions and Evaluating Answers
Once everything was set up, it was time to ask questions! NotebookLM suggests questions based on the entered information, but you can also type your own. For example, it suggested a question about Katie Tiger’s tribal enrollment number and the relationships listed in the 1966 oil & gas lease case.
I selected the question about Katie’s enrollment number. It explained that the Dawes Card said her number was 6517 but the Proof of Heirship stated 6516. It was right! I had incorrectly typed her number on the Dawes Card record. I also inquired about the relationships in the oil & gas lease case, and it correctly named all three children.
I wanted the complete ancestry of Felix Buster, which it should have through all four grandparents and five great-grandparents. Despite multiple queries, sometimes asking for his ancestry and other times for specific relationships, the responses were hit or miss. This shows that while the potential is promising, it’s currently not reliable.
Additional Features: Summaries and Key Topics
One feature I appreciate is the “summary” for each source. For instance, the summary for the 1966 oil & gas lease case states:
“This legal proceeding, published in 1966, is a notice that Felix Tiger, Louis Tiger, and Elizabeth Beaver, descendants of George Tiger, are requesting court approval to grant an oil and gas lease for their land in McIntosh County, Oklahoma. The lease terms include a 5-year duration and a $5 per acre bonus to G.R. Stirman. A court hearing is scheduled for March 21, 1966, and competitive bids will be considered at that time.”
Sharing Notebooks and Premade Examples
You can share these notebooks, a feature I might use to collaborate with my friend. However, I’d advise them to verify any answers provided by NotebookLM.
If you’re interested in experimenting with this tool, Google provides four premade notebooks to explore, including one titled “Introduction to NotebookLM.”
Conclusion: A Promising Tool with Room for Collaboration and Improvement
While NotebookLM shows a lot of potential, it’s clear that there’s still room for improvement before it becomes a go-to resource in genealogy research. If you’ve tried NotebookLM, I’d love to hear about your experiences. Are you familiar with any similar tools that might work better? Please share your insights and any other tools you’ve found useful. Let’s collaborate as we discover new tools to help with our research!
2 Comments
Leave your reply.