RLP 322: Review of “Co-Intelligence: Living and Working With AI”

Nicole (1s):
This is Research Like a Pro episode 322 Review of “Co-Intelligence: Living and Working With AI”. 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.

Nicole (41s):
Let’s go. Today’s episode is sponsored by Newspapers.com. Hello, welcome to Research Like a Pro.

Diana (47s):
Hi Nicole. How are you doing today?

Nicole (50s):
I’m great. It’s been fun listening to all of the lectures from the National Genealogical Society Conference.

Diana (57s):
I agree. It’s been always fun to hear somebody else’s case studies and their thoughts on researching and records. So a couple of my favorites so far have been Nancy Peter’s lecture on a German case study, which was very interesting, and then James Beidler’s session on newspaper research. Both of those had so many interesting things to think about in terms of our research,

Nicole (1m 22s):
Right. The newspaper lecture by James Beidler was really eye-opening for me. I think a lot of the things I had known, but I really appreciated hearing it all in one hour long lecture and to think about the ideas of gathering up the titles of newspapers for your locality, putting them in a locality guide. I just kind of got thinking about better ways to keep track of newspapers for research projects. So it was great.

Diana (1m 51s):
It was a reminder that there are so many newspapers out there and we have the big sites like Genealogy Bank and Newspapers Archive, Newspapers.com, but there are all these smaller groups like libraries that have microfilmed their newspapers years ago and maybe are now getting them digitized or maybe they’re still just on microfilm. So I remember when I was doing my accreditation project contacting a Texas library, just a small one, and they searched the specific issue that was mentioning my family for the death of our Dora Algae Royston. I knew when she died, And I knew specifically when it should be mentioning her death, but they only had a few of those issues for that week.

Diana (2m 38s):
And so I’m guessing that if her death was reported, it was maybe in one of the missing issues, but it was great to be able to hone in on a very specific time in place and a newspaper that I knew the family had been mentioned in it for other things and have that search done.

Nicole (2m 55s):
Yeah, that’s such a great example. For announcements, we wanted to share that our Airtable Research Logs for Genealogy PDF guide has a bonus file that is called Text Generation with Airtable AI. So if you purchase that PDF for $10, you’ll get a little bonus file that talks about how to use Airtable AI tools. Our next Research Like a Pro webinar is September 21st, Proving the Parents of John G Winn: a 19th Century New England Study by Karen Ramon, and it’s about how an undocumented history gave clues to John G.

Nicole (3m 36s):
Winn’s parents. Using the Research Like a Pro method, those clues turned into evidence! The development of a strong locality guide and a journey through courthouses and cemeteries, combined with online sources and a town clerk’s office, led to various records that helped conclusively document John’s connection to his parents. This focuses on Massachusetts, vital records, census, newspapers, cemetery, and probate records. So if you would like to join us for that, please register for the 2024 Webinar series and if you’re already registered, make sure you attend that and watch the recording.

Nicole (4m 21s):
Well, the next Research Like a Pro with DNA study group will be February of 2025. So be thinking if you want to join that and what project you might want to be working on for your research objective and be thinking if you’d like to be a peer group leader and apply on our website, you just need to submit one of your DNA research reports for us to look at. Some upcoming conferences include the Association of Professional Genealogist PMC Conference, September 19th through the 21st. It’s virtual and I’ll be speaking there about Canva and then the East Coast Genetic Genealogy Conference, October 4th through 6th in Maryland and virtually, and we’re both speaking there and the Texas State Genealogical Society Family History Conference, november 1st and second. It’s virtual and Diana’s giving several presentations.

Diana (4m 58s):
Let’s get to our topic for the day, which is talking all about a really great book that both of us have read and that has some very interesting thoughts about working with artificial intelligence. I wanted to learn more about this whole idea of artificial intelligence and using it for genealogy. And when I took Steve Little’s course at the GRIP Genealogy Institute this summer, he recommended Ethan Mollick’s 2024 book, Co Intelligence Living and Working With AI and I really enjoyed this book because it didn’t get too deep into the technical details. It was perfect for my level of understanding and it had a lot of examples, and this one is not specifically to genealogy, but of course we can draw our own conclusions and make our own connections there.

Diana (5m 50s):
It also has a lot of funny little sayings and thoughts in it. So I appreciated the humor and he gives us a lot of food for thought. I thought that because this was book was very recent, very new, published in the spring of 2024, that it would be pretty up to date with telling us what was going on with the large language models. And that’s a challenge because AI is continually changing and being updated. But the thing I liked about Co-Intelligence was that Ethan Mollick gave us some broad principles, and even with updated AI systems, those will still be valid.

Diana (6m 30s):
He is an educator and he has embraced the use of AI with the students at Wharton, where he is a professor of management specializing in entrepreneurship and innovation. And in his introduction, he writes, after a few hours of using generative AI systems, there will come a moment when you realize that large language models LLMs the new form of AI that powers services like ChatGPT, don’t act like you expect a computer to act. Instead, they act more like a person. It dawns on you that you are interacting with something new, something alien, and that things are about to change. So I love that introduction and that just gives you a little sample of his writing, which is very engaging.

Nicole (7m 15s):
It is engaging, and it was fun to review all of his ideas before we taught our Research Like a Pro with AI workshop. And it was really helpful to use some of the different ideas that he had to help everyone understand how you can use AI in your work. Well, with Co-Intelligence, it’s divided into two parts, and the first part of the book is a brief history of AI and how to use it. He walks us through the beginnings of AI in 1950 when Alan Turing was doing work in theoretical computer science, and then after that there was further development of AI, but it had ups and downs until 2010 when we saw AI being used in businesses, but not by the average user.

Nicole (8m 8s):
You’ll think of Google Search or Amazon with all their logistics. Then in 2017, there was a new technology announced by researchers at Google called the Transformer Architecture. And when that was developed, there was a huge shift. This eventually gave rise to the large language model like GPT, which could predict the next token a word or a part of a word. Then a huge breakthrough for us as consumers was open AI’s introduction of ChatGPT in November, 2022. Because it had a simpler user interface, we can now easily experiment with AI where AI had been in the background for our work as genealogists with indexing efforts from Ancestry and FamilySearch, we could now use it ourselves, and millions of people have signed on to ChatGPT and asked it to create poems, write essays, and much more.

Nicole (9m 1s):
When we first started using large language models, we did the same thing. Let’s make a poem or a song with it. Just fun things, but we weren’t really using it for genealogy right away,

Diana (9m 11s):
Right? I remember using it to write, I think it was a treasure hunt or you know, just funny uses that you’d think of, and then you would just kind of let it go by the wayside because you weren’t specifically using it for work. And I also used it at the very beginning for helping me to write some titles for presentations, which at the beginning they were so flowery and just used so many big words that didn’t even sound like me. And I had to do so much editing on those and now it’s getting so much better. But it is fun to think back to what we were doing right at the very beginning. Well, let’s have a word from our sponsor Newspapers.com. Today’s episode is sponsored by Newspapers.com, your go-to resource for unlocking the stories of your ancestors.

Diana (9m 55s):
Dive into the newspapers where your family’s history unfolds as you search nearly a billion pages in seconds. Newspapers.com offers an unparalleled treasure trove of historical newspapers providing a window into the past with papers from the 17th century to today. Newspapers.com is the largest online newspaper archive. It’s a gold mine for anyone seeking to uncover stories from the past. Whether you’re a seasoned genealogist or just starting your journey, Newspapers.com makes it easy to search for obituaries, birth announcements, and the everyday stories that have shaped your family. That’s like having a time machine at your fingertips. And here’s the best part, our listeners get an exclusive offer. Use promo code Family Locket for 20% discount on your subscription.

Diana (10m 38s):
That’s Family Locket at Newspapers.com. Sign up today at Newspapers.com and embark on a journey of discovery. All right, so let’s get to chapter three, where Ethan Mollick discussed four rules of working with AI and I thought this was particularly helpful. It only takes a few weird answers to realize that chatting with the LLM is more complicated than you initially thought. So rule one is to always invite AI to the table. And I thought this was so interesting and so good. This also was emphasized by Steve Little in our course to whenever we’re doing a task, ask ourselves, can AI help us with this task?

Diana (11m 23s):
And so in the book, Mollick emphasizes the need to experiment. We have to actually try this out, see what we can actually do with AI and what will be useful to our workflow or just to our lives. And this will really help us to understand its limitations, abilities and the subtle nuances in working with this tool. And I think that idea of subtle nuances means our prompts so important. You know, what we tell it to do is of utmost importance. We sometimes think it will just read our minds, but no, we have to get our thoughts onto paper, actually into the prompt box in order to see what we can do with AI.

Diana (12m 4s):
So rather than get frustrated when we don’t get the response we expected, we can use that. We’re trying to learn the AI strengths and weaknesses. So for us as genealogists, as we’re going through our research work, you know, whatever we’re doing for the day, whenever we have a task we can ask ourselves if AI can help. So tedious tasks like transcriptions can be done very well by AI, but it is up to us to discover how accurate that transcription is. And Mollick describes this as the jagged frontier, meaning that we don’t really know how well a particular LLM will work with a task until we actually try it as technology improves something we couldn’t do yesterday could very well be possible tomorrow.

Diana (12m 49s):
And that that kind of goes along with number two, be the human in the loop. So Mollick makes some really interesting observations about working with AI, and he says, first of all, it’s more concerned with pleasing you than being accurate. And this is why AI will often hallucinate or give a plausible answer that’s completely wrong. And I have seen this so many times. So for instance, I was working on a widow’s pension application. And I asked the LLM to summarize the details. And I thought it looked so great there. The answer was all right there with the summary. But then I checked it against the actual document and saw that every single date was wrong, and then I had to prompt it to use only the dates on the document, which then resulted in an accurate summary.

Diana (13m 34s):
So we really can’t assume anything when working with AI. We are the human in the loop. We provide crucial oversight, our unique perspective, our critical thinking skills, our human values and our ethics, and all of those are so important when we’re working in our genealogy and our family history work. So the more I work with AI, the more I see how important my training and my experience is. The LLM is simply using its training of data sets to predict the next word. And although it’s very, very good at this, it doesn’t have my thousand plus hours of analyzing records to make genealogical conclusions.

Nicole (14m 15s):
That’s true, it doesn’t. And I think sometimes we think of the chat bot or the large language model as being smarter than us, and there are things that it knows that we don’t know, but there’s areas of experience where we’re an expert and it’s not an expert. So we have to recognize those strengths and weaknesses. Alright, well the third role of working with AI is to treat AI like a person. And Mollick says that since AI is eager to please us, we need to tell it what kind of persona it should take. So in that way we would treat it like a person as far as telling it what kind of person it’s, so if we are doing research, we can tell it you are an expert genealogist that is clear and specific and will hopeful hopefully guide it to act appropriately when helping us with the task, giving it.

Nicole (15m 5s):
This persona helps it to activate the correct words and phrases and tokens associated with professional genealogists. So we can start all of our gen genealogy related prompts by giving a large language model this persona, And, I like that idea. I think that Mollick is wise to share that advice, especially because forcing us to think about the persona that we want the AI to take on helps us to think of how we can make a better prompt. So I think it’s good for the large language model and good for us. While his fourth rule of working with AI is to assume this is the worst AI you’ll ever use.

Nicole (15m 47s):
This speaks to the ever advancing world of AI. As you’ve all seen, the major companies are competing to provide the best model possible, and it’s exciting to see each new model that we can then experiment with to try new tasks. So if the current model can’t handle doing what you want it to do, perhaps the next model will handle it with ease. And as we’ve been experimenting with using AI for different tasks within the Research Like a Pro process, we’ve had some successes and some failures. So it’ll be interesting to see which of those failures can be made successes with the next model that comes out.

Diana (16m 29s):
Right. I thought that was a funny title for his number four, Assume this is the Worst AI you’ll ever use, but it’s so true when you think about it that it’s only going to get better and better. Well, in part two, after giving us those four sections of four rules of working with AI, Mollick goes on to give us several ways to consider AI in our work. And these were really helpful. And as Nicole, you said in our workshop, we tried to really note on our slides when we were using AI as one of these roles because it really helps us to think how are we using it now.

Diana (17m 10s):
So the first one is using AI as a person and Mollick stresses that AI is unpredictable something that we are all seeing. And in our institute course, when I took that with Steve Little, he would give us prompts and we all tried inputting the same prompts into say, ChatGPT or Claude, and no one received the same answer. So interesting to see. We’d put our answers in the chat and they were all different. So this is much different than using a calculator that will always give you the same answer to a problem. So since a doesn’t act like software, treat it like a human, understand the strengths and weaknesses of each model.

Diana (17m 52s):
So currently the ones that we are using right now that are popular ChatGPT-4o, Claude 3.5 sonnet, and Gemini are the ones that I have been working with the most. And I have tried having each one perform the same task with the exact same prompt. And it is fascinating to see how they differ. And I like to think of this idea, you know, of AI as a person, if I had five people in the room and And I asked them a question, I would get five different answers. And so the models are very similar to that idea.

Nicole (18m 27s):
The next role is AI as a creative, and this is where Ethan Mollick discusses hallucinations, which we’ve all heard of, and they kind of scare us. This is where the model makes up answers. And Mollick says that this may come from the training data, which may introduce biases into the answers. He also warns against small hallucinations such as little things that are inferred or slightly inaccurate or it just figured it out, but we didn’t really give it that information. So we’re kind of wondering how it got to that conclusion. So it can be tricky. We need to really work on fact checking it and making sure that we are aware of the hallucinations.

Nicole (19m 12s):
But the creative side of these models is also a good thing for projects where we need to get ideas, brainstorm things, make a creative writing work sample like a poem or a song or a fiction. So if that’s what you’re working on, you can get ideas for that. And brainstorming with AI is really helpful. I’ve done it a lot, And, I. Think it’s nice to have an assistant where when you’re brainstorming it can kind of regurgitate back to you what you’ve just said to it and then maybe add some other ideas to that. Maybe we just need a new way of thinking about a research question. You could ask the model to propose ideas to solve your question.

Nicole (19m 53s):
Maybe it will give you a new perspective. And AI is trained to process a huge amount of information, but we have to know how to work with it to ask the right questions. So maybe think about times when you need a little creative help and how you could get that help from a model, a large language model,

Diana (20m 12s):
Right? We don’t often think of creativity as part of our research, right? We’re working with facts, but there are definitely times where we need ideas. Well, another role Mollick gives us is AI as a coworker, And, I think this is one of the things that will be most helpful for us as genealogists. Anytime we have something that’s a mundane or a tedious task, we can ask ourselves if AI could do this quicker. And Mollick suggests we divide our tasks into just me tasks and me and AI tasks. So I really love that idea. In my world as a professional genealogist, a just me task would be analyzing the research and writing the report, making the appropriate connections.

Diana (20m 59s):
And, I say, that’s just me. Because even though AI could help write that report, my brain works best when I am writing and when I’m taking the information and putting it into words, that really helps me to think through what’s going on in a specific scenario in the research. So, you know, for me it would, it’s better for me to write the report myself, but a me and AI task would be if I took that same report when it was completed and asked AI to create a summary in a bulleted list. I’ve already written it, I’ve made all the conclusions. Now I can have AI even perhaps write the conclusion as well as the summary that I put at the beginning of the report.

Diana (21m 46s):
So I think it’s just wise to think about that. What is something uniquely we do and what would be good to have AI help us with?

Nicole (21m 56s):
The next role is AI as a tutor. Our go-to tool for learning something new usually is Google or YouTube videos. But you can use a large language model to learn how to do something and often the response seems quite good, but may need some tweaking. For example, writing an email. Maybe you want a quick list of steps for sharing Ancestry DNA results. When you’re inviting someone to share their results with you, we can ask the large language model, it’ll give us a nice bulleted list, but then if you try out the steps, some steps are missing and may need to be adjusted. So it is a bit of an experiment.

Nicole (22m 37s):
And using AI as a tutor for all kinds of technology related tasks or learning about various topics can be really helpful when you’re taking a class. One really great use for AI is helping with creating digestible notes from the class. And sometimes you may have access to the class transcripts to turn into a handout or notes. And so you can take that audio transcription of the transcript of the class and ask a model to summarize that or create class notes or turn it into something a little more usable. There is no end to how AI can help us learn.

Nicole (23m 17s):
We can use it to help summarize difficult articles that are above our understanding level or how to interpret a document if we don’t know about the historical context and the response may be incredibly helpful or not, but I think it’s definitely worth a shot. When I’ve asked for help with all these kinds of tasks, I’ve always been pleasantly surprised with the response and the help. If we are asking for historical context, it can be really a good place to get ideas. And then what I like to do is go find an actual source that I can cite for that idea by doing a Google search for that topic. But once it’s given me that historical context, I now have the words to go do the Google search.

Nicole (23m 59s):
So it’s a really helpful way to learn about something,

Diana (24m 3s):
Right? One example that just happened recently, we had a client project and it was a deed that had just a couple of letters that were abbreviations for something at the beginning of it. And the researcher, and I both looked at it and we didn’t know what that meant. So I took a screenshot of it, I put it in into Claude or ChatGPT, one of them, and said, you are an expert genealogist and you’re transcribing this deed. What do these two letters mean? And it used, it’s, you know, it’s training data to pop back right away with the response that was spot on. And it gave us exactly what those letters meant, which now I can’t remember what they are.

Diana (24m 44s):
And then it gave a whole paragraph an explanation of how that was used in deeds back in the day. So, you know, I couldn’t have really done that with Google because maybe there would be an answer there somewhere if you weighed through all the responses. But that was a such a good use as a tutor to help us learn something that we’re stuck on. Well, another role is AI as a coach. And this discusses the concept of expertise. Can AI help us become an expert in our field? So in my opinion, I learn and do best when I get feedback. And you may be the same way. So if you’re writing your very first genealogy research report, you have no idea if it is good or not.

Diana (25m 30s):
And I, remember my first report that I wrote, or a page or two was for my accreditation study group, And I turned it in, and then I got feedback. I think I had 20 things that I got feedback on that I could do better. And I was so grateful because I had no idea if I knew what I was doing or not. And then the reviewer then said, you’re really good at this, you, you should keep going for sure. So that was great, getting feedback from an expert. Well, you may not have access to somebody who’s an expert that you want to give your report to, but you could give that to the large language model and try to get some feedback. So if used appropriately, AI could really help us in many areas of our work as genealogists.

Diana (26m 16s):
I had worked out a report on my Clemsy Cline project. I uploaded that and I asked the model to suggest sections that needed further discussion. And I was really happy with the suggestions that were spot on that I needed to develop this idea a little bit or that idea a little bit better. So it was such a great way to get feedback from “an expert Genealogist”. Another thing we could do is, is give it a sample, you know, of a really well done report and then put our report in and say, okay, how could mine be better using this expert report as something to to judge mine against.

Diana (26m 58s):
So, you know, we’re often just working in our own corner as genealogists and a large languish model could become a coach for us.

Nicole (27m 7s):
Absolutely. Well, the last role of AI from Ethan Mollick’s book is AI as Our Future. And he talks about four scenarios for the future and what each could mean for us as humanity. The four scenarios are as good as it gets slow growth, exponential growth, and the machine God. So you’ll have to read the book to learn more about those, but it is important to think about the future and AI and how it will change our lives and how it’ll change our genealogy research. I think we’ve just begun to scratch the surface and what is certain is that change is coming. And if you haven’t already noticed the changes, then maybe you haven’t been looking very hard, but there’s AI incorporated into so many of the things we’re doing now and next week on the next episode, we’re going to talk about how it’s being incorporated into one of the tools we use a lot for our DNA research Lucidchart.

Nicole (28m 5s):
And it’s just interesting to see how AI is being included in many of the tools we already have. And I’m sure that you’ve seen that. Well, we both highly recommend Co-Intelligence as a good book to ease your way into using AI into all areas of your life, and particularly in becoming a better genealogist and family historian. So, good luck in your research efforts and we hope that you will try to learn a little bit more about this AI thing.

Diana (28m 37s):
Yes, I think sometimes we’re resistant to big changes like this. You know, we have done research this way and it’s worked for us all of our lives, so why do we have to embrace something new and different? But hopefully this episode has given you some ideas and Nicole And, I have fully embraced using AI and will continue to give ideas in in podcast episodes and in our writing because I think it’s so important to see how we can become better with the tools we’re given. So thanks for listening and we’ll talk to you next time. Bye-Bye everyone.

Nicole (29m 11s):
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.