Using Artificial Intelligence Tools to Expand Your Genealogical Research Universe

Hello and welcome to the My Heritage webinar series. I'm Maram Pier Lou, your host, broadcasting to you live from Eastern Massachusetts. Today, we have James Tanner, who is live with us in Utah for his class, "Using Artificial Intelligence Tools to Expand Your Genealogical Research Universe." Thanks to James, and thanks to the over 2200 of you for registering for today's live webinar. So, wherever and whenever you are, glad to have you with us.

And now, to introduce our speaker. James Tanner has a BA in Spanish and an MA in Linguistics from the University of Utah. He also has a JD in Law from Arizona State University. He spent 39 years as an Arizona trial attorney. James has over 42 years of experience in genealogical research. He is an avid blogger at Genealogy Star. He has been a family history volunteer for 20-plus years. Presently, he serves at the BYU Library Family History Center. James is on the board of directors of the Family History Guide Association. He is also fluent in Spanish and has done extensive Spanish genealogical research. Please put your virtual hands together and let's give James Tanner a nice, warm webinar welcome. How are you, James?

Welcome to the webinar.

Well, thank you. I'm glad to be here today. Let me switch over to your screen. That's the important part, and that looks great. Time is all yours.

Okay. Well, today we're going to talk about using artificial intelligence tools to expand your genealogical research universe. And, uh, one of the things I like to point out is that we are talking about the images. And in part of this, and this is an image that, uh, was generated by generative AI. And, uh, this is the kind of images that are being produced, um, outside, a little bit of outside of the, uh, realm of genealogy, obviously. But, uh, just so you know in advance, that a lot of these images are now being generated. The advantage, of course, is that certain rulings by the Supreme Court of the United States indicate that AI-generated images cannot be copyright protected because they weren't created by a person. And so, from that standpoint, they're great for me because I don't, uh, I need to find images that aren't, that aren't copyright protected in order to, uh, use them in my presentation. So now I can generate those images in a lot of cases.

So let's get along to it. Um, uh, I would point out that a lot of the genealogical community, the greater genealogical community, is pretty well still stuck in the paper-based world. Um, I still see, uh, when I go to the Family Search Library in Salt Lake City, I still see people coming into the library with their rolling briefcases or rolling suitcases full of documents and books and records and things. And, uh, I'm on the other side of that spectrum because my equipment that I have all into the library, if I can go and if it's permitted, obviously, is my phone, and that's about it. Unless I'm really think that I'm going to do some typing, and then I might carry in a laptop. So there's, uh, really a complete change in the way that I did work for over many, many years, and now I'm doing work completely online.

So I guess the main question we have is, are you ready for an AI future? Is there something, uh, you know, is this going to take perhaps a restructuring of the way that you're going to be doing and using and and taking advantage of information? So all I can say is from the very beginning, and I'll probably repeat a number of times, is that, um, you are, uh, that is AI is a general term that really isn't very appropriate, but it will have a great, has had an impact, a great impact on genealogy already, and will have even a greater impact, probably an accelerated impact over the next couple of years. So you're not going to find that things are getting any less complicated in this, in the sense of having changes and having ways of of approaching data change.

Um, one of the things that we need to always remember, and this is, uh, a quote from Arthur C. Clarke, uh, back in 1962, and he said that any sufficiently advanced technology is indistinguishable from magic. And some of the things that we're seeing today with generative AI, uh, appear to be magic. They just are beyond comprehension of things that you can't believe they can actually do that. Um, and the, and the way that the AI is impacting my particular way of doing research is equally as revolutionary because I am now using the chatbot. In this case, the one I use primarily right now is, is, uh, Microsoft's Copilot. And that basically is, uh, supplementing and and supplanting all of the information that I've been gaining previously through Google searches. Um, I find that to be a much more efficient way to work, and that's why that it's doing, it's actually doing that.

So one of the things that you may not understand or or realize if you're not keeping track of, of kind of the background that goes on with the larger genealogical websites is that this is, genealogy has been used, and the, I mean, genealogical websites have been using AI for well over 10 years. Um, and a lot of the developments and the things that you see on the major websites are automatic, are are basically based on AI programming. And they're the things that have, because of developments and and the fact that they're going on kind of in the background and and not making front-page news anytime very quickly after they're done, there, nobody would even know that they had happened or that that was the possibility of what was going on. And so the main thing that AI has been used about now is to answer questions about historical events, places, dates, and genealogy resources.

So you, if, when you're interacting with the AI, uh, you have two, two major parts of how they're put together, how the programs are put together. One is a big database, which they call a large language model. And that is a way of teaching AI how to use the language and how language is structured. And then you have the actual mechanism or the program that uses that, that, uh, information from the large language model and communicates back and forth. It's sort of the mouthpiece of the, the AI. And they're called chat or chatbots. And the chatbots, uh, can answer questions instead of just simply giving you a response that says, here's a website that might have an answer to your question. It gives you an answer. And then, uh, if it's a very sophisticated chatbot, it will also give you a footnote to the source where that information came from. It'll link it and tell you, well, this is where I got the information. The one that's doing that right now is Microsoft's Copilot. And, uh, right now, if you ask Copilot a question, it'll give you a, a straight answer, but it will also tell you either if it doesn't know, it'll say, I don't know. But if it does know, it will give you the information plus footnotes that show you exactly where the information came from.

So answering questions has become a, a almost constant, uh, background to the things that I do. And now I can get that information much more quickly and much more concisely from the chatbots. Now, another one is to translate text and speech from one language to another. Now, we need to do that a lot. It depends on how far back, obviously, you've gone with your, your genealogical research. But it's, uh, pretty, um, common that people will trace some of their lines or all of their lines back to a place that does not speak their native language they're using today. So if you're living primarily in an English environment, you may end up going into an environment that's Spanish or German or any number of other languages. And, uh, now the ability to translate and use the information from one language to another is becoming very, very transparent. Unfortunately, the one-hour presentation that we have here today is not going to allow me to get into a lot of detail on how to do all this. But the questions, if you had a question about how to do it, the best thing to do is to go to the AI chatbot and ask that question. Say, "How do I do this?" And, uh, and you'll get an explanation. So it's kind of like having a, a friend there that knows a lot more than you do, and you can simply ask a question whenever it occurs.

Uh, one of the things that's happened that's quite dramatically was kind of, was principally initiated by MyHeritage, and that was the ability to enhance and colorize old photos, animate faces, and, uh, and then give also, involved with that, of course, is giving ethnic origins based on DNA tests. All of the analysis of DNA has been used. When we talk about artificial intelligence, it's a huge umbrella to hundreds of pages of different types of programs that are being developed. And under that umbrella of AI is, uh, the analysis of DNA. That's being heavily impacting some of what goes on in genealogy today. It also is being used to index records and suggest record hints and create, uh, family trees. Uh, there are programs today that were, are computer-generated family trees. You might not, you may think, well, that doesn't sound like it's going to be very accurate. They can't possibly be accurate. But if you had one that was, uh, generating a family tree and then giving you all of the sources where they, uh, where that information came from at the same time, then I think you would probably, uh, begin to understand that that was, uh, could be something very, very useful. Um, and it will also match DNA results with other people. The matching programs, all of those are higher-level, uh, computer programs, and they can be able to create things like family histories or or biography or whatever of a person from all of the facts that it can glean online and the sources that are available.

Okay, but I think the major one, uh, program development that will impact, directly impact genealogy beginning about two or three years ago until from now on, is going to be the ability that they have reached sort of a high level of of ability to, to read handwriting with a computer. Uh, there are some, I call them semi-commercial type programs online that will do the same thing for an individual. But right now, the computer, the level of computer involvement is such that it isn't readily available for individuals unless they get into a paywall kind of situation where they're paying for transcriptions. But I would be guessing that within a very, very short time, meaning in the next year or two at the maximum, that there will be programs that will allow a, a fairly substantial ability of doing handwriting recognition and being able to be used generally by genealogists to do research. Right now, it's very memory-intensive and does not practical for individual computers. That's the problem. They're going to be sitting up on the, what we call the internet. And of course, there's going to be, uh, an i for people to supply that kind of information and dedicate that computer power. There's going to be a, a charge, but we'll see how that develops and and goes over the time.

So here's, for example, here's an example of what I can do with a document. So let's look at this document. It's, um, uh, this is basically a, from Spain, and it's a Spanish document. And if you, if you don't read Spanish or you don't, uh, have any idea what the words mean here, it, uh, could be kind of an obstacle for you to do any kind of research. Well, so what would be the help here if I were able to take this image and tell the, the computer to, um, translate all of this, all the text, maybe not all the handwriting yet, but all the text at least to, to English, so I knew exactly what it was saying? Well, that's exactly what's going on. And so here's an attempt by Google Gemini to read that same text. This is that same document, and it's given me a list of the translations of all of the different fields that are showing up here on this index. And it happens to be, this particular handwriting happens to be from Spain. It's Málaga, which is a province and municipal birth records. And this one came from MyHeritage.com. So this is a record that was on MyHeritage, and I pulled that off. And, uh, and so this was the transcription.

So what happened here with the transcription? Now, part of what I do, and I may repeat this a couple of times to emphasize, is that in order to test if a particular AI chatbot or an AI program, uh, is going to be helpful to me, if it's going to work, if it's, and to find out what kind of expectations I can have that that it's accurate or that it's doing its job the way it needs to do. And I, and what I do is very simple. I, I start by asking questions that I know the answer to. I use people in my, uh, lineage of my pedigree going back, my great-grandfather, particularly as a subject, because why? Because I know out of my head, I don't even have to go look it up. I know the information about him. And so when I'm doing a test, if it doesn't come up with the correct information, I know that there's some limitations here. And in this particular case, um, I had Google look at this, and it said that the next word down was "Milagro" and it was translated as "Miracle." Well, that's not correct. Um, actually, that is the word on the document is Málaga, and that's the province. Uh, apparently Gemini could, did not realize that Málaga was a place, not a word to be translated. So there's, these are the kinds of limitations that you're going to see as you work with this. But, but given the fact that I've got the rest of the information appears to be, uh, reliable, then there's incentive to use that for other purposes.

Um, it would be more in this case, since I speak Spanish fluently, I have no trouble using the, I mean, I don't actually need the document to be translated because of obvious reasons. But on the other hand, uh, if I move into a language that it's not familiar with, such as Hungarian or some other Eastern European language, I may have to rely more than I would, uh, recently that I would on, on a language like Spanish or German or French, um, and finding the information that I need. So I test the accuracy of AI with information I can already translate or already know. And I think this is a good model for, for what I would suggest to any new user. Is start out doing AI searches and asking questions you know the answer to, and you'll begin to see if the program actually responds in a way that is, that you think is useful. And then go to branch for something you don't know, and then see if you can verify that what it gave you is the correct answer.

In any event, uh, realizing that a computer program is based on, on what it knows. In other words, whatever's been put into its database is, is what's available. And, uh, you, and from the standpoint of genealogy and and having sources and looking at the original documents and things like that, it's all the same rules apply. And so you basically need to verify and, uh, be well aware of the fact that AI may be telling you just something to fill in and doesn't really have the correct answer. And by the way, is not flagging it and telling you that this is not been read sufficiently correctly.

So what else? Well, okay. So one of the things that I, I absolutely need is to understand more about the records themselves and the places where these records were created, created. Why is that the case? Well, whenever you're working in genealogy, you work from the known, what you know, to what you don't know. And, um, if I'm working in a particular, uh, place in in Europe or in a place where I have not, where I'm not just really, really acquainted with, like a town or a province that I haven't particularly visited, I may want to know more about that. I may want to know what the parameters are. When, when were the first churches established? When did the records become available? What kinds of records could I expect to see in that particular area? And these are all things that that, uh, the AI chatbots can give you, uh, provide help and give you, in a sense, a leg up in starting out with this area. So it will basically tell you what the rules of the game are, and it will, and help you understand that more quickly.

Now, when we say this, you know, the answer could be easily say, well, I could go out and ask the same question of a Google search. So why do I care whether I can work with an AI program? Well, the difference is what was produced here, uh, in response to my question, which was, "Tell me about the Santiago Parish in Málaga, Spain." And so it told me everything about the Málaga Parish, and, uh, that came out in about 15 or 20 seconds worth of waiting for it. And so I have just now jumped around. Uh, now, how do I know this information is correct? Well, we're not, we're not going anywhere with that question because, uh, uh, we have the same issues with every time we do a search or anytime we go anywhere. I mean, if I go into the library, uh, like the Family Search Library in Salt Lake City, or might be, or BYU Library, and I pull a book off the shelf, how do I know whether the book is accurate or not? How do I know the whole book wasn't made up? I mean, I really don't. In other words, we live in a world of of possibilities, but we don't, some of those possibilities include the the problem that what is being produced, what we see, is not particularly accurate, uh, and or is completely inaccurate. And so the whole process here is one of of making sure that that information is reinforced.

And so here we are. This is, I'm particularly specifically using, uh, Gemini here on the, for, for Google's program rather than Microsoft's program, Copilot. And the reason is because when I was doing the translation portion of that, it's Google has been using Google Translate for years, and that's their very, very sophisticated program. And as a matter of fact, the way that you use Translate, you may not have noticed lately, but if you go to Google Translate, you'll see that there's a, uh, a way to do documents and files and translate a document or a file. Okay.

So I would, uh, um, just, just as a matter of note, that the genealogical community, the companies are beginning to implement AI search apps. So you're beginning to see that the drift in that direction. Now, uh, MyHeritage introduced a, its AI, uh, record search program in December of 2023 at RootsTech. And then I reinforced that at RootsTech and indicated that this program, so that we now have that, that capability using MyHeritage to, to, uh, do searches in the database of the records that are available from MyHeritage. And the advantages, of course, is that that you're asking questions and getting responses rather than just was getting a lookup of of a number of cases or whatever.

So this is the program that came out from MyHeritage. It's called the AI Record Finder program. You have the choice of making two different, of of involving two different, um, uh, styles, you want to might say, of of working with the program. A conversational style that's casual and formal. I suggest when you're going to go try one of these types of things that you try all the different iterations and options that are available, um, and just to see what you, which serves you the best and which reflects of how you'd like to get the information. I mean, I would, this is for me, personal, obviously, and that would be that I would always opt for formal over anything casual because I'm looking for for specific information about specific items, not, uh, uh, in this case, record finders for the individuals I'm researching in, in my family tree, rather than any just sort of casual conversation kind of things, like, "What kind of, what, what would you recommend for lunch?" or something like that.

Uh, unfortunately, by the way, this is an observation that, uh, much of what's been being broadcast and and discussed online in the news about AI searches has focused on things that are are tremendously trivial and have no, are not a real challenge to anybody doing research. So the, the real question still comes, can you do what I would call real research on, on AI? And the answer is yes, if you're willing to spend the time to focus on and asking questions of things that you specifically want to know the answer to. And if you do that, if you, if you get out of the general, kind of, "What word, what would you recommend I go eat lunch?" or something. My question is, why are you needing that kind of help from your computer? I don't know. But anyway, those seem to be the level of what most people, most of the discussion about AI is being, is carrying on.

So right now, this is a part of, of MyHeritage. If you have access to MyHeritage, uh, go up, uh, in the menu and and do a search, uh, of the Record Finder, AI Record Finder search. And then, uh, uh, see what kinds of things it produces. That's the, the, the experience that you have when you start to use AI is going to be very much personal compared to, let's say, a Google search. And the reason that for that is because the AI is responding to your, your inquiries, your prompts, your, your way of formulating the question. And you can see if you try changing the question a little bit, you'll see that it will, the program will respond differently. So there's, uh, there is an interaction portion of the AI experience that is, um, far different than doing a strict search by for a name and an index or for, uh, some website from Google.

So this is, this is kind of the core of the things that are happening in, uh, that are going to make a big difference. Um, and if you, for example, if you were asking about a specific kind of record, and you would then could say, "Can you, do you have any records that have, that show taxes, tax records that show on maps?" Anything you think of from an AI standpoint, you may end up with the AI chatbot coming back and saying, "I don't know. I don't have that information. I don't understand your question. Please rephrase it." Whatever. But, uh, that is the, the interactive with the, with the interacting like that with the computer is the way that you learn how to phrase and how to talk and communicate.

Uh, here's another opinion as you're doing this, and that is that as you are working with, uh, an AI chatbot, the, the computer on the other side with this program and this language ability, and, you know, there's some people have have indicated that it'd be a good idea to think, think like it's another person and you're talking to another person. My answer, I don't think that's very wise. What I would suggest is thinking like it's a computer, which it is, and then trying to figure out what its parameters are and what kinds of things it can and won't or won't answer. So this is, uh, it's a, it's a matter of a learned skill of learning how to use the words to, to, uh, get a desired response from the AI.

So here's, uh, here's what, um, you can search. And this is, uh, using the AI search on, on the menu. You can see the menu bar over there on the, on the right. It says AI Record Finder is the, uh, first item on this about the collection catalog on the right-hand side of the screen. And so then when you check that Record Finder, uh, you need to put in some information, obviously. And as I was, as I indicated pretty, uh, much in the last, uh, the last few minutes, I said that I use information that I already know. In other words, H, let's ask the, let's ask the computer to give me something that, that I already know is there and see if it knows how to find it. So that way, I, uh, began the process by putting in, uh, the prompt, which is up at the top in that little green bar. I wrote in, "Tell me about Harold Morgan, born 1892, died 1963." This is my grandfather. So obviously, I knew the core things about my grandfather. Um, and this way, um, one of the things that you can build on as you're using this information is, uh, assuming that you're getting the correct, because Harold Morgan was correct. But the Harold Francis Morgan who lived in Boston or was born in Boston was not my person. My person obviously came from Utah. And so, uh, you're still in the situation that you have previously. This has not gone away. In other words, uh, even though this, the AI chatbot for program is sophisticated and responds to, uh, what you would call plain text or plain natural language, what kind of language you're using to ask questions, it isn't necessarily anymore, uh, make any fewer times of inclusive mistakes where you include where it includes other information that's similar, uh, and may match, but is not the same person or the same, uh, information that you're looking for.

So the response was this, and it, and in fact, it gave me quite a bit of response of records that I would be, that I would find useful. And in this case, but it also said, and what came down at the bottom of the screen here is, "To refine the search results, provide additional details such as locations, names of close relatives, or other relevant life events associated with." This always jumps you into, which comes first, uh, the chicken or the egg, the cart or the horse, or whatever. But what you're talking about here is that as you feed it more detailed information, it will then provide you back more detailed information. And you're wondering, well, if, if I already know all this information, then why would I be asking all these questions of it? And the answer is, uh, the function of it is is not just to answer questions, but in this case, it's a record finder. So it's going to be looking for records that apply to you as your particular individuals, um, what you're wanting from that particular individual, how you want to, what, what information you need.

So I kind of scrolled over to get more records and, uh, found even more information about, uh, my, uh, my grandfather. And in this case, I found a California death record from, uh, 1963, which says that he died in Los Angeles, California. Um, this is not a record that I had known, seen before. Not, not that I would have not have found it had I been doing some more research. It's just the fact that, um, this was, is just not a record that I would have, uh, necessarily have gone after and looked for.

So what I found as I continued to do this was that, um, I was actually getting responses back from from supplying a little bit more information. So I gave them my birth and the place and everything, and then the chatbot became more responsive. And, uh, in this case, it gave quite a bit of information. And it gave me a second option at this point on MyHeritage to create an AI biography. And so that was another way to, uh, expand on this. Um, I have tried that, uh, on some of my ancestors. And of course, what you get out of the, uh, AI biography is is limited to what you actually, what the type of information that you actually have. In other words, it's, if it isn't already in the program, if it hasn't been entered, and you don't have some documents and things about the person, it can't make it up from thin air. It, it has to have some basis. But, uh, and part of the attraction of of creating a biography, for example, is that it does it very quickly and does it very thoroughly and and, uh, can contain information that has a one-to-one relationship with the records that have been, uh, that have been identified.

So it's kind of a suggestion is to, um, um, is to try to build on from from actually using the information to build a, a greater ability to find more information because it's just, it's, it's incremental in that way. And so what's happening here is as you react with the Record Finder, you can see records are going to start to show up that you previously had not discovered. And if you, and in this case, there's a big red "Save the Record." And the answer to that is, save the records, keep the record, create a larger, uh, basis for making information about any specific person. And so, you know, if, if there's, there's a little bit of resistance here when you start getting into things like creating a, a biography. And the resistance is, is how cure, how, uh, accurate will this be? Will it be anybody there that I mean, is it just going to throw in a bunch of stuff that you really don't need? And, uh, the answer is, well, try it and see what happens. Um, it may surprise you one way or the other. You may be surprised that that there is so much information that you didn't weren't aware of. And then on the other hand, you may be surprised that there's not any information and it can't make up anything. So, uh, it just really depends on the on the circumstances. And that's that's nothing new. That's not like it's a problem that was created by AI or was not expected to happen because anytime that you're dealing with historical facts, and particularly genealogically oriented, uh, historical facts, you're going to find information that you're going to find that some information is missing, that there just isn't a correct, there's no records that are available in whatever repository you're working from, uh, to create that or get that information. And, uh, then you hopefully you don't experience what, uh, may happen on occasion, and that is that that AI, because it gets frustrated, decides to make it all up and give you inaccurate information.

Well, yeah, and that could happen. Uh, I can, I can probably, that type of thing was overblown at the initial, with the initial introduction of generative AI. And I would think that all of the chatbot people who are now working on their programs have have tried to avoid that. I know with Google, uh, and, uh, and Microsoft, both are very much involved in that. So that's going to be the one of the questions that is going to be answered.

So this is another program that was introduced back in December by MyHeritage, and that is the AI Biographer. And it would then take, like I just was was highlighting, it will take all the information that you have and turn it into a structured biography. Um, you may not like it, you may like it, you may think it's wonderful, you may think it's not wonderful. But even if it isn't, it's probably a, uh, gives you a good first draft, if you will, or a basis for, uh, expanding on the, on the, uh, biography and saving you some time and getting the documents in the basic structure organized.

So this is, uh, this is these are all part of, uh, of MyHeritage. Uh, and so basically, there are the, some of the tools that have been coming that are extremely valuable in in giving us information. I'm not saying that MyHeritage is exclusive on this, but these particular tools have been, uh, are, uh, innovatively enough of an innovation and enough way of working, uh, that is, I think they're unequal out there in the, in the greater genealogical community. And here's my admonition: it's a good idea to attach all the pending record matches and record hints before you create the biography. Now, you notice I use the word here "record hints." And if you have watched presentations that occurred in RootsTech, 2024, this last about a month or so ago, you will realize that MyHeritage now has a set of record hints as well as record matches. So there are actually record hints on, on MyHeritage now.

Okay, so here's, uh, um, another, uh, search that I did. The reason why I did this search was because I wanted to know if, this was about another one of my great-grandfathers who were, who, uh, lived most of his life in Utah, was born in California, but I wanted to know if there were any, what kind of of information there might be in another product that was introduced by MyHeritage called OldNews.com. It's a separate website. So if you haven't, if you're not aware of that, that, uh, they have created this website that partially overlaps some of the record, the newspaper records that are already on, on, uh, MyHeritage. But MyHeritage intends to use OldNews.com to be the, uh, basic repository for all of the records. They're adding from the countries around the world as they acquire newspaper collections. And so this is an ongoing project that was, uh, new from, basically again, from RootsTech.

So the question was, did I want to have any, do they have any information about my great-grandfather, my great, my great-great-grandfather in this case, uh, who was, uh, lived in a little town in, in Central Utah called Beaver, Utah? There's lots of other states, places called Beaver in the United States. And, uh, in this case, it turned up a, uh, obituary in an old newspaper for December 6th of 1895. And the date of the article was December 7th, 1895. And that was that, uh, this particular, my particular grandfather had passed away, and they wrote a nice little article about him and about his his life.

So I guess what I'm saying is that, um, we have, we have some new tools to do research, and we have some new offerings from, at least from MyHeritage, in the case, in this case, that are fulfilling those and giving us additional information that we may not have seen or even guessed was available previously. So now, um, this is the question, and his name was Sydney Tanner, and he was born in 1809 in Greenwich, Washington State, Washington County, New York. And he died in Beaver, Utah. And so I get that information. This is the, uh, the page of the search engine that I got for, for the search for Sydney Tanner. Now, I didn't ask specifically for newspapers. I just said I was looking doing a search. And, uh, by the way, the record matches and record hints that are appearing on, uh, on MyHeritage are also AI-generated. So, you know, if you, if you were kind of under the impression that AI was something that just sort of blossomed about a year ago and that it's all a big deal now, no, it's been a big deal now for years and years and years, and it's being used and has been used now for well over 10 years by all of, as I mentioned earlier, by all of the major genealogy programs.

So here's a, uh, another obituary. Uh, and then we, when you get to that point, when you've added your record matches and your record hints and all that information in, then you're probably ready to, uh, to do an AI biography. At the time that I was playing around with the program and getting this information, uh, it was still sending a notice to you by email that the biography was complete and they had sent me a copy of the what they had created. Think you can imagine it takes a tremendous amount of computer power to do this on a regular basis, all day long, every day, 24 hours a day, around the world with all the millions of people who want information immediately.

So another problem, this goes back a few, few to a few years. MyHeritage introduced an AI-driven image enhancing app. And that was to enhance. It's called MyHeritage Photo Enhancer. And it was, uh, the start of a series of programs that became highly sophisticated and used probably some of the most impressive generative AI sort of things that are going on. In the first case, um, this is, uh, using the AI. I put on the photo on the left. That's the photo that I started with, and then I had it enhanced on the right. And so now we have the enhanced version on the right. Now, uh, a lot of what you see is up to the individual, uh, preferences and what they think are is, uh, as good or bad or or helpful or not helpful. And so I'm not saying in any instance instance here that I think that the photo on the right is better than the one on the left, or the one on the left should be kept and not the one on the right. It's too fake, or whatever you want to call it. But the answer is that if it, if it improves the connection with the photo and helps, uh, to show, possibly what the person looked like in real life, and that may be, uh, enough to justify doing this kind of a, an enhancement.

Now, if you're like me, and you have inherited some of your relatives' and hand and ancestors' photographic collections, you'll find that, uh, there are plenty of these photos out there that need to be, uh, repaired or, uh, at least looked into a little bit. Now, some of the, the, uh, information may be completely lost. So let's kind of move through this so you can see here that, uh, if I run this through, that I can, uh, uh, I can put it. I didn't get much change, but I can go into Photoshop, which is a very expensive Adobe Photoshop program, and make some more. But comparing these two is, is not even fair. It's not even un, it's unfair because the, the program, the, the, uh, image enhancement program is part of the MyHeritage, um, entire, entire collection.

So here's another example of a photo. So the one that, this is repaired by MyHeritage, and then when you take that, you can enhance it, and then when you go from that, you can colorize it. So that there's all these stages. Now, if I do that, and you don't like the results, of course, you don't have to keep the results. In my, I would prefer not to have the colorized version, and the black and white version seems to be, to me, to be more contemporary and more like. But perhaps in some cases, the colorized version is very much better. So I went in and used Adobe Photoshop to, to try to do the same thing, to repair the photo. And I, I think that there is, uh, very not significant amount and not to justify, for me, it's not to justify the time that it took me to do all that in Photoshop as compared to what I got from directly from MyHeritage very quickly.

So there's also photo tagging. That's an ability there to tag. And then we have animation that will animate a photo. Uh, it can be pretty scary sometimes. And then also, we have Deep Story. Okay. So basically, what we have are some fairly innovative things that are that are happening. So you can see, I think, think from this that, um, there's still, there's a lot of AI of of artificial intelligence that's already been well integrated into, uh, at least MyHeritage and and perhaps some of the other programs. But we're going to see new innovations and and fabulous things in the future because this is a, not a static discovery. This is something that's going to affect everything on down and through all of the, uh, the various areas of of genealogical research that we have going today.

Okay, well, thanks for watching. And I can assure you, there's a lot more to learn and a lot more to hear about. But, uh, we're kind of limited to an hour or so.

Thank you, James. Christine is asking, is is there going to be a part two for this webinar?

I guess it's up to whoever schedules these webinars. You could talk to Daniel about that.

All right. Sit tight for a few minutes. I have a bunch of questions. But let me just go through these, one or two or three announcements and door prizes. So our next, uh, MyHeritage webinar will be a closer look at the 1850-1950 US Census records with a little extra help by Diane Hendrick. That'll be, I think, on May 14th or 15th. So that's coming up. You can head over to, um, FamilyTreeWebinars.com/MyHeritageWebinars in order to register for that. And you can find all of the, uh, existing MyHeritage webinars in the Legacy Library. Here's a direct link for that, just slm/myheritage for the ones in the library. These are all free. Um, we had a bunch of them during the 24-hour marathon. So really, really good stuff. "Unlocking Birth Records: Exploring Birth Dates and Alternate Sources" by Lisa Listen. "What's New and Exciting in MyHeritage Historical Records" by Moo Kland. And searching your family in New York City: Resources and Techniques from Mike Mansfield. And I heard with this one that somebody actually brought down a brick wall during this webinar. So, oh, and we, we also had one from Michelle Leonard about clustering with the MyHeritage label. So these are just, of these four were from, uh, the 24-hour marathon that we had fairly recently, a week or so ago. James has a bunch of other webinars in the Legacy Library as well. And if you go to, uh, FamilyTreeWebinars.com/JamesTTanner, it will bring you to those directly, and you can check those out.

All right, we need to give away a door prize or two. This is the new database James was talking about. This is the new newspaper subscription that MyHeritage has. This was, uh, released, announced at RootsTech in, uh, March. And so this is their, their large newspaper database. I'm going to go to our live audience and I'm going to pick somebody for that. Bang. Let's see. That is going to go to Mazella Bon. So congratulations, Mazella. And I'm just jotting your name down so I can send you an email later and get you instructions for how to get your prize. And our next door prize. Oh, you know what? Huh. You know what? Let's go back. I'll just, we'll just do this. The two door prizes I have here. Obviously, I, I thought I was giving away a, a DNA kit, but I guess I don't have that slide. So we'll give away a MyHeritage Complete Plan. Somebody will really enjoy this. This will give you access to all the, uh, the family tree building and tree consistency checkers, everything like that, as well as all the, uh, historical records, uh, in at MyHeritage. So that is really great. And let's go. That is going to go to, let's see here. That's going to go to Denise Ireland. So congratulations, Denise. And we did that. Okay.

And of course, we'd like to encourage you to become a Legacy Family Tree Webinars member. This week, there's actually a half-price sale. So you can get it for $25. That gives you a whole year of access to Legacy Family Tree Webinars, over 2,000 classes in the library. All right, let's bring James back on and we will tackle some of these questions here.

All right, Joy wants to know, can AI translate documents from handwritten sources?

Well, there wasn't enough time to get into that in detail, but the answer is yes, it can. And handwriting recognition has been, uh, matured and is now a, a not, I would say commonly, but it is now being extensively used, uh, by larger websites to transcribe and translate handwritten documents. The question is whether that will ever filter down to where an individual user could be putting in a document and get it transcribed or translated. And, uh, I guess the observation I make is that that's still taking enough computer power that right now, uh, it's not quite feasible, but it will probably be very much used in the future. And I'm saying in the next year or so, I'm down the road, very far.

Okay, great, great. Nan's asking, how is MyHeritage AI Record Finder better than just doing a standard search from the research tab?

I, I don't think I have been done enough searches that I would be able to give an an educated answer. I got different responses from each. It's

Not necessarily better. It's just going to give you some things that, for me, were surprising that I wouldn't have been looking for. Is the difference mainly in the way that you can talk to it and interact with it? Like you can just type whatever you want into the record finder, you know, as if you're talking, and it'll give you answers as opposed to, as opposed to searching, you know, within a data set for a name. Yes, that's that's really the whole key to it that you're basically interacting with it. And as you give it more information or provide more details, then it will automatically go start looking for more specific items.

And I think it would work better if, if the problem, of course, is that I've been on, uh, for me judging is that I've been on MyHeritage for years and years and years and, um, I just have, like say, hundreds of, not thousands, tens of thousands of record hits sitting out there. And so even if, you know, if I got serious, I would have to spend full time trying to process all that information that I already have.

Okay, so, uh, Elise is asking, uh, also about AI Record Finder. Uh, she tried using it, but it came back with records from her own family tree on MyHeritage. So, is there a way to use Record Finder, and maybe you don't have any idea about this, but, um, she's wondering, can you do it and and exclude the stuff from your own tree? You know, just tell it, just tell it, "Oh, that's interesting." Well, that's obvious. Yeah. So, so Elise, what a great suggestion. I know about my family tree, please look elsewhere. Okay, so, yeah, or don't, don't include results from my family tree. That would be a really direct way of doing it. Have to use please and thank you, but it's also nice to do that. But so Elise, I hope you try that right now so you can post an update, um, and and let us know what. So if you put, "Don't include results from my family tree," um, let us know if it will give you any results. That'll be really interesting.

Uh, let's see. Willen wants to know, I don't know if you tried this or not, can you save and print the results of a chat with AI Record Finder? Um, well, whatever you get is on the screen, and it's not, um, necessarily a text file. I mean, it's not something if you wanted to get, get capture it, you could use screen, you know, a screen capture or whatever of the of the information. I don't think that it's, uh, I've noticed one thing, and they're very persistent. In other words, once I've done a search like that, it's sort of there forever. Like it, all it doesn't seem to go away. Diana is saying you can, uh, do copy all. Is the suggestion?

Uh, Elise is getting back to us on the Record Finder situation, and she said, uh, that that it, she tried that, and it didn't work. So what I would say then, uh, Elise, I would write to MyHeritage and explain the scenario and and suggest to them to make it work so that you can exclude your own tree in the results. So they might not have thought about that scenario, but if you bring it to their attention, then they can work from that. Um, so that would be what I would suggest. Yeah, I think that's a good way to do it.

Uh, let's see. Edward wants to know, is the information about Harold Morgan coming from actual records that have been reviewed and placed in this format, or is this from user-submitted trees, or both? I will add, that's both. Some of that was, some of those records that were coming up were from, uh, record matches, smart matches, I mean, where they were matching it to other trees. But that's that's what you get with MyHeritage. I mean, you're you're always going to get the combination of smart matches and and record matches. So it's it's always useful.

All right, um, one or two more questions. Uh, let's see. Linda is saying, if you have a profile for an individual in a program like MyHeritage, and you add some documents and newspaper clippings, and you ask for a biography of the person, will the bot include information from the articles or the documents, and not just records? I was seeing that it, that that what was produced was extremely detailed and very nice. I'm not sure exactly, I didn't look at it in from the sense of saying where every piece came from. It was just the one I, the one I did was very impressive. It was my grandfather, which I was very impressed with. Just immediately, it had the photographs, it had everything, and I just, I thought, wow, I didn't, I could have spent two hours or three hours doing this. So interesting. Well, I'm going to have to try it. I'm going to have to add, uh, I I definitely have a lot of newspaper clippings about my grandfather, so I'll have to add that and then give out give that AI biographer a try and see if it it includes them or not. That'll be that'll be fun. It seemed to be very detailed, however.

And let's see, oh, okay, so here's an interesting, uh, question from Dale. He, Dale says, is the information creating a new AI person's MyHeritage ID page, or does the information go to the existing ID page for that person? So is it creating a new page, new person page, or is it going to the existing person? No, no, no, no, no, no. It's just, it's using the existing. I don't see anything new there. Okay. All right, great.

Well, that is going to do it, uh, for today. James, this has been really, uh, thought-provoking and and certainly we had a great conversation even before the webinar started. So I'm going to be off checking out, uh, Copilot and all sorts of other things and checking out, um, more of this AI biographer and, wow, just so much, so much to play around with and adapt to our our ology. So this has been great. Thank you.

Okay, and we we'll look forward to having you back. Clearly, there's going to be a need for part two. You've got a lot to share. All right, uh, thanks to our audience for joining us today, and we hope we'll see you, uh, next time at our upcoming MyHeritage webinar with Diane Hendrix. See you everyone. Bye-bye.