RLP 329: Using AI in Report Writing

Nicole (1s):
This is Research Like a Pro episode 329 using AI with Report Writing. Welcome to Research Like a Pro a Genealogy Podcast about taking your research to the next level, hosted by Nicole Dyer and Diana Elder accredited genealogy professional. Diana and Nicole are the mother-daughter team at FamilyLocket.com and the authors of Research Like a Pro A Genealogist Guide. With Robin Wirthlin they also co-authored the companion volume, Research Like a Pro with DNA. Join Diana and Nicole as they discuss how to stay organized, make progress in their research and solve difficult cases. Let’s go.

Nicole (40s):
Today’s episode is sponsored by Newspapers.com. Hello everybody. Welcome to Research Like a Pro today.

Diana (48s):
Hi Nicole. How, are you doing?

Nicole (50s):
I’m doing really well. How about you? What have you been doing?

Diana (53s):
Oh, I’m just working on a couple of different projects. My case study on the Roystons and Cessnas and then also my study group project on the Weatherfords. so I usually like to just focus on one family at a time, but right now I’m kind of split between the two. So that’s fun though. What, are you doing?

Nicole (1m 14s):
Just juggling multiple projects as well. I’ve been trying to wrap up my Dyer research from the spring and I’m gonna continue with that for the study group if I ever wrap up the last report. And then of course I’m still working on my kinship determination project for certification for the portfolio I’m hoping to submit soon to BCG the Board for Certification of Genealogists. And mostly what I’ve been doing since I’ve been traveling a bit and on airplanes the last week, driving in cars and whatnot, whenever I have some time I’ve been pulling up all the photos I took from my research trip to Oklahoma and just in the description area for the photo, like the metadata, I’ve been putting in kind of like a very simplistic citation, just showing like Love County Oklahoma Tax Roll the year and like the person so that I can quickly see when I look at that image, what is on this tax roll or what is on this deed, like who’s this deed for.

Nicole (2m 19s):
And then using that and and copying and pasting that into my research log along with the link to that photo because I use Google Photos and so all of my photos on my phone are automatically backed up to Google Photos. Then I can use that as as if kind of like I’ve digitized that page in the tax roll And. so now I have a link to it. And so anytime I want to go back to the exact image from my research log, I can click on that link to Google Photos. So I’ve kinda got a little system down. so I processed a lot of the images and I’ve been just kind of correlating them and looking at all the different tax years that are covered. It’s a lot of the, the years from like 1908 to 1920, so there’s a couple ones missing, but I luckily was able to get most of them even though I needed to spend a couple more hours in the treasurer’s storage room, which was a jail behind the courthouse And, it is a jail cell.

Nicole (3m 16s):
So that was fun. Kind of eerie to be in there going through the old textbooks alone with all of the bugs and the dead crickets. But it was wonderful, you know, it was a genealogist’s dream to just have access to all of those, but it’s been really great to review those. And of course it’s taken me a lot of time because just life gets in the way and things happen and I can’t believe it was May that I did that and now here it is October and I’m still going through everything. You can digitize a lot of records quickly, but then processing them and putting them in the log just takes a while.

Diana (3m 47s):
Oh yeah. And didn’t you have about a hundred images?

Nicole (3m 51s):
Yes, over a hundred. Yeah.

Diana (3m 53s):
So that is a lot of work. A quick question about the tax books. Were they alphabetized? Was it fairly easy to find the families?

Nicole (4m 3s):
Well, until I realized how they were organized, it was a challenge. But then I finally realized some of the earlier years there was just one tax book per year, but then later they were divided into two tax books per year and they were two townships in each one. So then I had to check both of them because one of the ancestors lived in the city in a different township and then another ancestor I was looking into lived in a different township, so I had to check both and then they would have like the personal property and poll taxes separated from the property and real estate taxes. And so the real estate taxes were not alphabetized, those were organized by the location of the land.

Nicole (4m 47s):
So those ones were harder. And just to make sure I didn’t miss anything, I went through every page. So

Diana (4m 55s):
That’s a lot of work.

Nicole (4m 55s):
That was a challenge. Yeah,

Diana (4m 58s):
That is tricky. So interesting. It’s such a good tip that you really have to understand how things are organized. And it, it pays dividends to go look through the books and figure that out at the beginning. And it’s no different with a lot of our tax records that they often will call them captains districts, you know, in the ones I’ve used, and if you don’t realize that there could be multiple captains in districts that you have to look at each one per your person, you could completely miss your person And just thinking that the first one, you look at the alphabetized list and think, oh, they’re not there. They’re missing a net year, but actually they were on the next list. Tax records are wonderful, but you’ve gotta learn how to use them and it really does come with experience, doesn’t it?

Nicole (5m 43s):
Yeah. And so the first time you do it, you should just look through the whole entire set of everything for that whole year. Because even if you do find your person on one set, he could be also listed on another set of, you know, delinquent or somebody who paid another kind of tax. Like sometimes you’ll have a carriage tax list that’s just totally separate and so you, you gotta look at all of them.

Diana (6m 4s):
Right? Or sometimes at the end they’ll have all the people who paid taxes for property that are not living in the county. Yeah. And they’re listed separate and sometimes they’re mixed in. So it just really depends on the place and the time.

Nicole (6m 17s):
Yeah. Good tip.

Diana (6m 18s):
Well let’s do some announcements. We are excited about our Research Like a Pro Webinar Series for November because it will feature our very own Nicole Dyer and that will be Saturday November 16th at 11:00 AM Mountain. The title is Four Generations of the Elder Family: Verifying Documentary Research with DNA. So the topics for this webinar will be autosomal DNA, the Leeds Method, proving your pedigree, Shared cM Project, standard deviation, independent descent lines, targeted testing, Kentucky, Missouri, Oregon, and Washington. So Nicole will be showing us how she used DNA to prove all the documentary work on this specific line that starts with my husband.

Diana (7m 6s):
So I am looking forward to this. It will be a really great introduction to using DNA for something like this, for confirming a pedigree. Our next Research Like a Pro with DNA study group starts on February 5th in 2025 and registration will open for that December 1st. We have our peer group leader application on our website if you’re interested in a complimentary registration. And as always, join our newsletter for coupons and to find out the latest information about FamilyLocket and what we’re doing. We are looking forward to the Texas State Genealogical Society Family History Conference, which will be November 1st and 2nd and it’s totally virtual.

Diana (7m 51s):
You can join us from wherever you live. I will be presenting there and it will be a great conference with lots of classes focused on Texas and Southern research as well as methodology, DNA and AI.

Nicole (8m 9s):
Wonderful. Well for today’s topic, we are talking about Using AI in Report Writing. And as Genealogists, we are always looking for ways to enhance our research and streamline the process, especially writing and writing can be time consuming. Writing a research report can sometimes seem overwhelming, so using artificial intelligence tools can help us overcome writer’s block. This is the last episode in our series about how Diana used AI for a complete project about Isabella Weatherford. We have already talked about how she used AI to help with writing the objective, transcribing a lengthy pension document for the timeline, making a locality guide, creating a research plan, and then following the research plan and putting that into the research log.

Nicole (8m 56s):
So in this episode we’re going to talk about using AI to help write the report. And this is one place where AI really shines because writing and language that is its strength, these large language models. So before we get into the specifics of report writing, let’s talk about understanding the role AI can play in our work. Rather than viewing AI as a replacement for our expertise, we should see it as more of a collaborative tool. Throughout the research project with Isabella Weatherford, Diana found AI to be in valuable in several key roles. First coworker, these are from Ethan Mollick’s book, right?

Diana (9m 37s):
Right. Co-Intelligence.

Nicole (9m 38s):
That’s right, Co-Intelligence by Ethan, Mollick and I just love his book. It’s so interesting to think about all of these ways that AI can work with us. So the first one is a coworker, and when it comes to a research project, especially with writing, we like to think in terms of what writing can be done just by myself and what writing can be done with the help of AI. And so when the evidence analysis and correlation was written, that was something like just myself can do. But then AI can help with time consuming tasks like summarizing lengthy historical text and AI is really good at summarizing and that’s not something that I need to really do myself.

Nicole (10m 19s):
It’s okay to offload that kind of thing to AI. And then with the creative assistant role, sometimes we feel stuck or just need some new ideas or a fresh perspective. And this is another thing that AI can help us with. It can generate new ideas and approaches we haven’t considered. Another role is as a tutor, AI can help summarize complex information or just help us as we’re trying to understand a concept that we need to write about and get up to speed on historical things more quickly. AI can also act as a coach and sometimes we will give our writing to AI and ask for feedback and ask for ideas about how to improve the writing, look for inconsistencies and suggest areas for improvement in the writing.

Nicole (11m 7s):
So coworker, creative, assistant, tutor, and coach, all of these different roles the AI can play, can help us with our writing as Genealogists. We are definitely going to remain the experts guiding the research and analysis that we’re doing, but AI can support and enhance the process. We just have to remember that it’s not going to replace our critical thinking and decision making skills.

Diana (11m 29s):
Right and I love thinking of those different roles for AI. Well, let’s talk about a few of the challenges because you might be tempted just to use AI to completely write your report because it’s sometimes hard to write the report, but there are some challenges. For example, you’ll want to track how you use AI throughout the project so that you know when to cite its use in the report. And then of course we have to fact check each AI response to check for hallucinations. And then finally, we do not want to lose our unique writing style and tone and I have a very specific way I like to write my reports and how I like to phrase things.

Diana (12m 14s):
And so I was careful that it kept my tone and how I did the report. So as we go through the different pieces of the report, we’ll be sharing how I actually did that. First let’s have a word from our Sponsor. Today’s episode is sponsored by Newspapers.com. Break down genealogy brick walls with a subscription to the largest online newspaper archive. Did you know Newspapers.com has over 1 billion pages of digitized newspapers dating back to 1690. Their growing collection includes papers from the US, UK, Canada, Australia, and beyond. Discover birth and marriage announcements, obituaries and everyday stories about your ancestors in seconds.

Diana (13m 0s):
Newspapers.com can help you fill in the gaps between vital records and reveal details about your ancestors lives that you can’t find anywhere else. Their easy to use search feature lets you filter your results by date, location, specific paper and more. When you find something interesting, Newspapers.com makes it a snap to share it with family and friends. You can even save it directly to your Ancestry tree. Come explore 1 billion pages and make infinite discoveries today on Newspapers.com. Use promo code FamilyLocket for a 20% discount on your subscription.

Nicole (13m 37s):
Let’s break down how AI can assist in writing each section of a genealogy research report using the Isabella Weatherford report as an example. First we have the the objective and that informs the reader and reminds ourselves of the purpose of the research. For the Isabella Weatherford report, you used AI to refine the objective. So here’s how it worked. You provided the AI in this case, Claude with the research question, key identifiers for Isabella and a sample objective from a previous project. The AI then generated a polished objective statement. Here’s what it produced. The objective of this research phase is to examine the economic and social conditions in Dallas County, Texas in the early 1870s and their influence on Isabella D.

Nicole (14m 23s):
Weatherford’s life and marriage prospects. Isabella was born on fourth March, 1858 in Missouri. She first married John H. Carpenter on January 16th, 1874 in Dallas County, Texas. Then later married Robert Cisnie Royston on January 16th, 1877 in Van Zant County, Texas and died on May 9th, 1942 in Tucumcari County, New Mexico.

Diana (14m 47s):
Right? So it’s always really important to get that objective front and center. I like a very first in my report so that I will remember in case I come back to this report and I can’t quite remember what was the focus for this project. Well, next I like to put the limitations section and this part, I think it’s typically best just written by the researcher. So as Genealogists, we understand the constraints of our project, whether related to time, access to records or other factors. So for the Isabella Weatherford project report, I wrote the limitations myself. I wrote first that the project was limited to 20 hours for research and writing.

Diana (15m 29s):
Second, that artificial intelligence was used to create summaries for portions of the report. And that third, the research was limited to records available online. So I call this the just me approach. And it makes sure that the limitations are accurate and every project will be very specific about what those might be. I decided to put in the part about artificial intelligence using it because in case something had snuck into the summaries that wasn’t actually accurate, even though I tried to do fact checking, I wanted to remind myself that I did use them. After the limitations, I do a summary of results.

Diana (16m 11s):
And even though it’s at the very beginning, I write it last. And this is where AI can really shine because it is so good at summarizing existing text. So for the Isabella Weatherford report, I organized the research, I correlated information and evidence, and once I had all of that body of the report written, I added that text to Claude and asked that to generate a summary of the key results. And it produced a nice bulleted list and I included the idea to start each sentence or each portion with an active verb and and so it did. And so that was great.

Diana (16m 52s):
It said things like, Ttace the Weatherford family’s movements, discovered Isabella and her family, analyze Dallas County’s economic and social conditions, explored potential migration routes and examined contemporary accounts. So it actually did a really good job on the bulleted list and I liked the way that it wrote it up. I reviewed it to make sure it was accurate and I did make some edits. I constantly editing whatever AI writes because sometimes it will use words that I think don’t sound very natural or I’ll have it re redo it and try to do better.

Diana (17m 32s):
But this collaborative approach of me and AI worked really well and saved me time. So having this summary of results there for myself or anybody else reading the report, we’ll just give you an overview and help you know what’s coming. I feel like it really sets the stage because if it’s not your own research, it takes a while to wrap your mind around the writing of a research project. so I felt like the result summary can give you that map and then you can go ahead and really dig into the actual research.

Nicole (18m 8s):
Fantastic. The next section that appears in a research report is the background information or the summary of the starting point. This can actually be one of the most time consuming parts of a research report, especially if you want to try to include all of the previously known information, which you don’t have to, but sometimes you want to. Well, for the Isabella Weatherford project, AI helped streamline this process of writing up the starting point. And in the past there were three complete reports about Isabella’s mother, Clemsy Cline Weatherford. There were several timelines, a locality guide and a project document. So to you solve these documents to help write the background information. Diana began by setting up a project in Claude, and if you don’t know what a project in Claude is, it’s kind of like a custom GPT, but it allows you to up upload a bunch of documents, and you can also do this in Chat GPT, but we just really like using Claude for this.

Nicole (19m 1s):
And you can upload previous reports, timelines, objectives, just all of the research to date on this project. And you do need a paid account to use Claude Projects, but it’s such a valuable tool, we really like it. Well, all of these things that you’re uploading become the Project Knowledge And. it gives the AI a comprehensive data set to work with. Next Diana prompted Claude to generate background information based on the provided data. The AI produced a detailed narrative that traced Isabella’s family history, including her parents’ likely marriage around 1839 and Izard County, Arkansas, and the family’s movements between Arkansas, Missouri, and Texas over the following decades.

Nicole (19m 44s):
Then Diana carefully reviewed what AI wrote, making corrections and adding nuance where needed. For example, she had to refine some of the date ranges and added context about the family’s occupation and social status that the AI hadn’t captured this approach. That was very much a me and AI approach, saved so much time. It’s like having a great assistant who can do a lot of the legwork and then you come in and finish it up.

Diana (20m 7s):
This might be my very favorite part of using AI in a research report because the background information can be tough. I had three complete reports to try to synthesize, and there was so much repetition between the three because each one of them, you know, was kind of summarizing the previous one and I was not looking forward to trying to write that. And so I was really, really happy with how that worked out. Well, after you’ve got your background information completed, and it’s always good to get that done because it helps you really synthesize, remember everything that you started with. Then you work on the findings and analysis.

Diana (20m 49s):
And this can be the hardest to write just because you’re trying to organize everything from that research log into paragraphs with complete sentences. And so this is what I did. I had filled out my research log knowing I was going to use Airtable’s AI Assist feature. And so I had really carefully filled out things like the comments, the citation, the actual summaries, all sorts of things. so I used that to quickly create some paragraphs. And, it really saved a lot of time in transferring the information from my log to the report.

Diana (21m 30s):
So we haven’t talked a lot about Airtable Assist AI, but it is a paid feature and I would recommend that you perhaps fill out your research log thinking that you will use that, but then just pay for the month or the months that you’re actually writing your report because you can just pay for one month and then you can use it to write your report and then you can cancel it. So I deliberately wrote all my analysis in the comments field, so AI could use that in the report. Then after I had AI generate those basic paragraphs, I wrote all the transition statements and analytical comments myself.

Diana (22m 12s):
So this made sure that my voice wasn’t lost, my reasoning was there, everything was as I wanted it through the report. And as we talked about in the previous episode with the research logging, when I was using a county history, I really needed to use AI to help summarize it. And so I actually used Claude’s summary on that and I wrote a citation that the summary was by Claude, so I wrote Lisa Maxwell, Dallas County from TSHA, which is the Texas Handbook online.

Diana (22m 52s):
And so I put that physical information and I did a semicolon and wrote summary by Claude 3.5, sonnet 31 July, 2024. And so this is why it’s important as you’re doing your research to remember, put it in your research log or your project document what you have summarized and what you’ve had AI summarize so that you can sure you cite that correctly. And I was really glad I had done that. And, it was very helpful to be able to just quickly copy and paste that summary right into my report when I had finished up that main body of the report, then it’s time to do the conclusion.

Diana (23m 32s):
And here again is a time that AI can really shine. And I used the Me and AI approach here because I took all of the report that I had gotten formulated and organized and uploaded that to my Claude project. And then I asked Claude to draft the conclusion based on the full report. And I had a really specific prompt, that I wanted it to base it on the main points and then also to note how the report met the objective. And then of course I reviewed that and refined it. And then finally is the section called Future Research Suggestions. And I had made a point in my research log to put these ideas right in my research log as future research suggestions.

Diana (24m 19s):
So then I used Airtable’s AI Assist to generate a section of the report based on what I had put into my research log. So that was so helpful. I had things like explore the neighbors of the Weatherfords in the 1870 census, search for military records to see if Henderson fought for the Confederacy, and the AI Airtable Assist just organized my ideas into a nice list. And it sometimes suggested some additional ideas of research, which I agreed with, or if I didn’t, I didn’t use those. So I really appreciated the help that it was to have AI helped me with those future research suggestions.

Nicole (25m 3s):
Oh, I love that. It’s so nice to just be able to automate some of the parts of the writing the report that we really do want to have that future research suggestions section at the end. But it is tedious sometimes going back through and thinking about all the future research ideas we already talked about or if we haven’t talked about them yet, going back through and realizing what we need to do next. So it’s so nice to have those a little bit automated.

Diana (25m 29s):
Right.

Nicole (25m 29s):
Well, I’m gonna talk about some tips for integrating AI into report writing now. These are some tips that Diana came up with after writing her report for Isabella Weatherford. Use AI throughout the research process, not just in the writing phase. AI can assist with everything from formulating research questions to analyzing documents. We can just ask ourselves, how can AI help me with this specific task? And be creative in finding ways to use AI’s strengths. We can set up a research log with AI assistance in mind, include fields that will make it easier to generate report sections later on. When using AI, always note which model you used and the date. This can be crucial for proper citation and for reproducing results later if needed.

Nicole (26m 12s):
Although, we all know that large language models give you different answers every time. So sometimes you just can’t reproduce results, but you can try.

Diana (26m 20s):
That’s true.

Nicole (26m 21s):
Also, keeping links to specific AI chat sessions, this can be helpful if you need to revisit or refine AI generated content that you’re using outside of your research log like in Claude or Chat GPT. And it is nice that now some of them are searchable and you can search through your previous chats. You can make full use of AI project features like Claude Projects, upload all relevant project information, including previous reports, timelines and transcriptions. So helpful. Don’t be afraid to experiment with different AI models for different tasks. You might find that one excels at transcription and the other is better at summarization. Then lastly, use multiple AI models when gathering locality information. This can help you get a more comprehensive and balanced view of historical context.

Diana (27m 6s):
It was so fun to go through a whole project and see how AI could help with research and especially with writing. It certainly made my work easier. I think often we put off writing the report because it just sounds so hard, but it became really fun to see how AI could help And. it was especially valuable in that background information and getting something written, getting a first draft and I think sometimes we just need something to work with. And so having AI help with summarizing the county history, doing that result summary, the conclusion of future research suggestions just gave me so much for my work, got that report just really in good shape.

Diana (27m 52s):
So even though it might be tempting to see if AI can just write your entire report for you, and I guarantee it will try if you give it some information, but remember, it really is a tool and shouldn’t replace your expertise. So I really thought through as I was doing each step, what my role was and what AI’s role was. And of course, as we have talked about many times, we want to verify it because we are writing a report and we want it to be accurate. And if we’re doing client work and it’s a professional project, we need to be extra careful. And so I worked hard to make sure that my unique voice and my analysis, all my experience, were right there at the forefront of writing this report on Isabella.

Diana (28m 42s):
So I hope everyone listening has enjoyed this series of episodes about how I used AI and hopefully you’re seeing the possibilities, you know, whether you’re using AI as a coworker, a tutor, or a coach. I think as AI continues to evolve, it’s going to be so exciting to see how we can best use it to help us in our research.

Nicole (29m 6s):
Agreed. Well, thank you everyone for listening. We hope that you have a great week and we’ll talk to you again next week.

Diana (29m 14s):
All right, bye-Bye

Nicole (29m 14s):
Bye. Thank you for listening. We hope that something you heard today will help you make progress in your research. If you want to learn more, purchase our books, Research Like a Pro and Research Like a Pro with DNA on Amazon.com and other booksellers. You can also register for our online courses or study groups of the same names. Learn more at FamilyLocket.com/services. To share your progress and ask questions, join our private Facebook group by sending us your book receipt or joining our courses to get updates in your email inbox each Monday, subscribe to our newsletter at FamilyLocket.com/newsletter. Please subscribe, rate and review our podcast. We read each review and are so thankful for them. We hope you’ll start now to Research Like a Pro.