RootsTech 2024 Early Bird Pricing Ends November 19, 2023

 

Click here to register: https://www.familysearch.org/rootstech/

AI could help the FamilySearch.org new user experience

 

Over the years, I have probably helped a few thousand new users of the FamilySearch.org website. I have never viewed the website to be even slightly "user friendly." The recent addition of a new startup screen for those logging in the first time is a major first step towards rectifying this situation. 

However, the screen does not show again if your login in still in your computer's cache. You can compare this to the still in existence startup page that has been around for quite some time. 

Artificial Intelligence (AI) has recently been in the news. However, AI has been actively used by computer programmers for many years. The earliest successful AI program was written in 1951 by Christopher Strachey, later director of the Programming Research Group at the University of Oxford. See https://www.britannica.com/technology/artificial-intelligence/Alan-Turing-and-the-beginning-of-AI

Here is one of perhaps hundreds of quotes that could easily be found online about using AI to improve the user experience. 

AI has the potential to greatly improve website personalization, providing users with a more personalized and engaging online experience. By collecting and analyzing data about user behavior and preferences, AI algorithms can make personalized recommendations and tailor the website experience in real-time. See The Role of Artificial Intelligence in Website Personalization.

The FamilySearch.org website has a huge database of information. For example, the Research Wiki has over 100,000 helpful articles about how and where to find information for new and experienced users. When I talk with someone who is just starting out using the website, I almost always guide them to the Research Wiki. Almost uniformly, the new users are amazed, overjoyed, and mostly surprised that instructions about how to do genealogical research are free and readily available. When time permits and when someone shows interest in actually doing research, I also send them to The Family History Guide.  

The challenges of the Research Wiki and the rest of the FamilySearch.org website is that they are "static." The user must supply all the initiative to finding the resources in the website. My wife and I are tasked with teaching all the new missionary volunteers at the BYU Library Family History Center. We start with how to login to the website. Immersing the website in AI would allow new and experienced users to actually use the website without fighting with it. For example, I have to log onto the website sometimes dozens of times each day. Every time, I have to go through the process of logging on the same way - every time. What if when I opened the program, it recognized me (like by Subaru does)? What if the website then asked me if I wanted to return to the last page of the website I was using or wanted to do something else? What if there was also a way to opt out of the entire process and go directly to the Family Tree? What if the FamilySearch.org website tailored the website experience to my use of the website in real time?

Another issue with the FamilySearch.org website is the lack of internationalization. Some modest attempts a have been made so far, but the website is essentially my same English speaking experience even when I am looking at in Spanish and usually it is even sometime harder to use when I elect to view the website in Spanish, for example. AI is quickly becoming a way to translate text faster and more accurately. Why is the Research Wiki in Spanish a weak experience compared to the English version? 

Let's suppose that the website not only asked me what I wanted to do, but had a way for me to start a conversation about my research goals. This expertise to help new and experienced users exists with the pool of genealogists such as those at the BYU Library Family History Center who have years of experience. Why not transfer this pool of experience to a website. Guess what, that is already available and done. That is essentially what is the basis for The Family History Guide website. By the way, The Family History Guide website is now linked directly from FamilySearch.org. The Family History Guide is in essence the basis for an AI entry experience with the FamilySearch.org website. 

I will probably have a lot more to say about this subject in the near future. Stay tuned.

The New Improved FamilySearch.org Startup Page

Above is a screenshot of the vastly improved FamilySearch.org startup page. You can see from the startup page that a real effort has been made to engage new users of the program. If you sign into the account you will get your own personalized page but if you are a new user you can scroll down on the startup page to find out substantial information about the website. If you are registered, you will get the older startup page after you sign in. 

Here is an additional screenshot of the next scroll screen.

The links on the additional scroll pages provide navigational access to the website. In this screenshot, you will learn about the archive of historical documents that are available on the website. 

The next screenshot shows the next scroll down section of the startup page.

Here is the last segment of the startup page. 

I guess my comment would be, its about time. 

Lynne M. Jackson to be RootsTech 2024 Keynote Speaker

 

From an email notice from RootsTech 2024:

RootsTech by FamilySearch is honored to announce its first keynote speaker, president and founder of the Dred Scott Foundation and great-great-granddaughter of Dred and Harriet Scott, Lynne M. Jackson.

A remarkable woman, Lynne Jackson will take the RootsTech main stage on Friday, March 1, 2024, to speak on the importance of remembering and connecting with ancestors, touching upon the story of her great-great-grandparents, Dred and Harriet Scott, and how their legacy has shaped her life.

Also remember:

 Click this link to register: https://www.familysearch.org/rootstech/

Artificial Intelligence: Is the cat out of the bag?

 

By PawełMM - Own work, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=76953571

The full question here includes whether the bag was already empty. There seems to be an invalid assumption that the current developments in artificial intelligence are somehow revolutionary rather than evolutionary. If you or anyone understands and knows about the history of the development of computers and programming over the past 100 years or so, you would not be surprised by the current developments or unduly concerned. Yes, there are concerns, but these concerns are also not "new." They are merely repeating what has been said about industrialization from the time it began in about 1830. Here is a short history of the basis for the evolutionary developments in AI. The problem lies with the people who use the information.

The basis for the ability of artificial intelligence to create text and images comes from large language models (LLM). Here is a short summary of the history of LLMs from Bing, “A Summary of the History of Large Language Models”, 2023. Retrieved from Bing on October 19, 2023. Note the list of sources provided by the search. 

Large language models (LLMs) are neural networks that can process and generate natural language using massive amounts of data and computational resources. They have evolved from the early attempts to create rule-based systems that could mimic human conversation, such as Eliza in the 1960s, to the modern models that can perform a wide range of tasks, such as GPT-4 and Google Bard. 

The development of LLMs has been driven by several key innovations in natural language processing (NLP), such as:

• The introduction of Long Short-Term Memory (LSTM) networks in 1997, which enabled deeper and more complex neural networks that could handle longer sequences of data.
• The creation of Stanford’s CoreNLP suite in 2010, which provided a set of tools and algorithms for various NLP tasks, such as sentiment analysis and named entity recognition.
• The emergence of Transformer architectures in 2017, which improved the efficiency and performance of neural networks by using attention mechanisms to focus on relevant parts of the input and output.
• The adoption of self-supervised and semi-supervised learning methods, which allowed LLMs to learn from large amounts of unlabeled text data without requiring human annotations.

LLMs have demonstrated remarkable abilities to understand and generate human-like text, as well as to acquire general knowledge and facts from the data they are trained on. However, they also face several challenges and limitations, such as:

• The ethical and social implications of using LLMs for various purposes, such as misinformation, bias, privacy, and accountability.
• The scalability and efficiency issues of training and deploying LLMs, which require huge amounts of data, parameters, and computational power.
• The generalization and robustness problems of LLMs, which may fail to handle novel or out-of-distribution inputs, or to explain their reasoning and decisions.

LLMs are one of the most impressive and influential achievements in AI technology, but they are also constantly evolving and improving. The future of LLMs may involve new architectures, algorithms, datasets, applications, and challenges that will push the boundaries of natural language understanding and generation.

Here is a list of sources and citations to sources for the summary of the history of large language models:

• A Comprehensive Overview of Large Language Models by Humza Naveed, Asad Ullah Khan, Shi Qiu, Muhammad Saqib, Saeed Anwar, Muhammad Usman, Naveed Akhtar, Nick Barnes, Ajmal Mian. This is a review article that covers a broad range of topics related to LLMs, such as architectures, datasets, benchmarks, efficiency, and challenges. You can cite this source as follows:

Naveed, H., Khan, A. U., Qiu, S., Saqib, M., Anwar, S., Usman, M., Akhtar, N., Barnes, N., & Mian, A. (2023). A Comprehensive Overview of Large Language Models. arXiv preprint arXiv:2307.06435.

• Large Language Models 101: History, Evolution and Future by Scribble Data. This is a blog post that provides a brief introduction to LLMs and their history, evolution, and future. You can cite this source as follows:

Scribble Data. (2023). Large Language Models 101: History, Evolution and Future. Retrieved from Scribble Data on October 19, 2023.

• Large language models: their history, capabilities and limitations by Snorkel AI. This is a web page that explains what LLMs are, how they work, and what are their strengths and weaknesses. You can cite this source as follows:

Snorkel AI. (2023). Large language models: their history, capabilities and limitations. Retrieved from Snorkel AI on October 19, 2023.

• Studying Large Language Model Generalization with Randomized Training Data by Yuchen Zhang and Percy Liang. This is a research paper that investigates how LLMs generalize to unseen data by using randomized training data. You can cite this source as follows:

Zhang, Y., & Liang, P. (2023). Studying Large Language Model Generalization with Randomized Training Data. arXiv preprint arXiv:2308.03296. 

None of these developments could have happened without the initial development of super fast computers, huge memory storage capabilities, and the internet. Which came first, artificial intelligence or computers? The concept of AI came from the earlier concept of thinking machines. The earliest idea of a "thinking machine" came in the 1830s when British mathematician Charles Babbage envisioned what he called the analytical engine. Viewed in the context of history, AI as it exists today was inevitable. 

What does all this mean? Essentially, the current notoriety of AI is based on developments that started more than a hundred years ago. The current handwringing and predictions about the end of the world, have been going on since before Karel Čapek's novel R.U.R., which introduced the word robot in 1921, and can be glimpsed in Mary Shelley's Frankenstein (published in 1818). See Wikipedia: AI Takeover.

What will happen to genealogy as soon as one genealogy company works out the details of using AI to analyze the information in their data base and family trees? You can see a glimmer of what is already happening with the suggestions now being made when you add a new ancestral line to an Ancestry.com family tree with their record hints and suggestions for parents. With the constant and accelerating development of AI programs, it is certain that how we do genealogy today will be different tomorrow.

Challenges of the FamilySearch.org Family Tree, Now and in the Future

For whatever reasons, both the FamilySearch.org Family Tree and the entire website face some serious challenges now and in the future. These challenges can be divided into two separate but related general categories: technological changes and data related issues. For the purpose of this post, I am not including issues that arise solely in the context of the temple ordinances performed by members of The Church of Jesus Christ of Latter-day Saints. 

The first and most serious challenge is data related and can be summarized by the old computer admonition: "garbage in - garbage out." The issue is how to prevent the Family Tree from having so many unsupported, inaccurate, and duplicated entries that it becomes so unreliable and full of errors as to be unusable. This issue arises in the dichotomy between encouraging new users to enter their basic family information and the need to put some reasonable controls on both the format and content of all "new" entries. Behind this particular issue is the ever-present problem of duplication of effort which I will explain next. 

Genealogical duplication occurs at two levels; when a new individual is added when that individual is already present in the Family Tree and when research is done by those who do not use the Family Tree to determine if the information they are researching is already available and documented in the Family Tree. Let me give an example of each of these duplication issues. 

The most common cause of duplication in the Family Tree occurs when a person who is unaware of or ignoring possible duplicates adds a name to a family or adds an entire family that is already recorded in the Family Tree. In many cases the new duplicate entry lacks supporting information such as dates and places. Because of the lack of complete information, the FamilySearch search program will not identify the new entry as a duplicate. Of course, the website can look for duplicates but the system as it now exists, often fails to "see" that the newly added individual or family is a duplicate entry until some additional information about the new individual or family is added by other users' research. 

Here is one way this duplication can occur. Let's suppose I add a name such as "John Smith" from my own personal records with limited supporting information such as that he was born in "about 1800" in the "United States." There is an good possibility that the "John Smith" I am entering will be a duplicate, but there is no way for either the person entering the information or for the computer program to determine which of the thousands of John Smiths are the duplicate or duplicates. When this happens, the program can offer possible duplicates when the user submits the limited information, but because the user does not know who their person is and cannot match the name to an existing entry, the user likely chooses to create a new person. This works fine if the user goes on to do additional research, finds the duplicate or duplicates and merges the entry. However, this is not the case when the user does not know how to do the subsequent research or is ignorant or avoiding the duplicate possibility.  Unfortunately, the website is designed to accept vague entries such as the about 1800 and United States entries in my example above. 

From my own experience, this problem of initial duplication is extensive in Latin America and other areas where the Family Tree has a large number of "new" users who are adding information about their immediate ancestors but ether choosing to ignore the suggested duplicates because they don't know what to do about them or because they think that they are creating "their own" family tree. This issue can be resolved to some extent by education as I will explain below.

Duplication becomes a more serious issue when the person entering "new" information is extracting individuals or adding families from census or other records without systematically verifying family connections. A prime example of this is the early extraction program in England where baptism, marriages, and burial records were individually extracted and showed up as duplicate individuals in the Family Tree; with three or more for each person entered. These duplicates are still being found regularly by researchers. What is not surprising about these duplicates is not only are they common, but ongoing individual and institutional extraction programs are currently adding hundreds of thousands of duplicates. 

Another example of the wholesale addition of duplicates comes from allowing old and new GEDCOM data to be added directly to the Family Tree. There are some people who deny that this is happening but experienced researchers who are watching their own entries find this occurring regularly. Those who are adding the entries do not look for duplicates and assume that they can add their "own" information to the Family Tree. Currently, the process for adding entries to the Family Tree from a GEDCOM file require the user to review whether or not the entries are duplicates but some users ignore the process and mark all their entries as new and thereby flood the Family Tree with up to thousands of duplicate entries. 

I could go on practically indefinitely about the duplicate issue, but I think that I have given enough examples to illustrate the problem. This brings up one of the other major issues and one that contributes to the duplicate issue which I mentioned above. This is the issue of entry level training or education. Although the FamilySearch offers several different pathways to learning about the Family Tree, users can always choose to skip the training and start adding names directly. There are presently no requirements to learn anything all all about the website before entering information into the Family Tree. You can enter names into the Family Tree with nothing more than a name. The website will note that dates and places are missing, but still allows the name to be entered. Why does FamilySearch resist the need to train people how to use the website before making entries?

I must digress here to explain why I use adding "just a name." This occurs when I am entering information from another research source such as Ancestry.com. I need a "place holder" in FamilySearch so that I can immediately start transferring information I already have with sources in my Ancestry.com family tree. Any name I add is always connected to a family where I am doing on-going research.

Back to training. There is no lack of training available. Again, referring to my extensive experience in helping new Spanish speaking users from Latin America and around the world, I find that they cannot use the website simply because they lack some really basic information about how to use it. Once I explain the relationship between records and entries and how to find the records they are relieved to know what to do.  The lack of available and required training is the one biggest obstacle to these new users having a discovery experience. Over the years of working on and helping to develop websites, I have found that adding some "Getting Started" buttons does not work when the user is supposed to know how to properly enter names, dates, and places. Warning messages that you haven't done a certain task correctly are useless unless the website provides the information to properly enter the information. Failing to have some introductory information and notice of the standards for entering information guarantees garbage in and garbage out. 

What else? The technology that is called artificial intelligence has recently progressed to the point where the FamilySearch.org website is simply old and out of date. There is no reason now that new and experienced users could not enter valid information using a conversational interface. The website should be asking users what they want to do (and adding an option for experienced users to opt out of everything except data entry and correction.) This technology already exists. The technology to construct family trees from valid sources with more accuracy that almost all potential users also exists but there is apparently a perceived FamilySearch problem that this will end up cutting out the user in the data entry experience. It is interesting that worrying about a new user entering his or her personal family information takes precedence over the accuracy of the entire website. It is possible for anyone beginning to use the website to find that information about their dead ancestors is already in the Family Tree. I commonly find that for Spanish speaking users who are struggling to find an ancestor, someone else in their family has already entered the information they are looking for and, as illustrated above, the new user is duplicating the research. The Family Tree is supposed to be universal so why should any new user be forced to rediscover their part of the universal family tree by doing duplicate research? An AI interface could help this new user have a good experience without later finding out all the work had already been done. 

Moving on, record hints are helpful but new users, without training, do not know how or why to use them. With an AI interface, the website could simply say, "I see you have record hints, do you need help in adding sources to your part of the Family Tree?" Why not have the website itself help to keep the information accurate instead of leaving it to experienced genealogists to waste their research time correcting the bad entries of others. 

The issue of researchers who duplicate research that is already in the Family Tree occurs both through lack of knowledge about the existence of the FamilySearch.org Family Tree and specific avoidance of using the Family Tree because of its existing reputation for inaccuracy and duplication. One fear of using artificial intelligence is that people will be replaced and lose their jobs. Those of us who are now spending an inordinate amount of time maintaining and correcting the Family Tree do not need this job. We would gladly turn it over to AI unless, of course, FamilySearch wants to start paying us to maintain the Family Tree then we might also worry about losing our jobs. 

I have a lot more to say about these issues and will probably keep writing until I pass on to whatever reward I get for doing all this work in the first place. 

Survey on Training Before Using the FamilySearch Family Tree

 

My friend, Kathryn Grant, is taking a survey. You may know Kathryn from all the videos she has done on the BYU Library Family History Center YouTube Channel. Here is a quote of her explanation about the survey. 

There are two camps of thought about teaching people to use FamilySearch Family Tree:

* Ask new users to take some brief training so they understand how to use Family Tree.

* Don't ask them to complete any training because it will make them not want to use Family Tree.

Where do you fall? Share your opinion on this brief survey: 

 https://docs.google.com/.../1XoFe7HcoNAcoia1QVkmL0I3PBP3.../  

This survey is NOT being conducted by FamilySearch, but it is being done with their knowledge and approval (as in they said it's okay for me to send the survey 🙂 ).

Please feel free to pass this link on to others. I'd like to get as many responses as possible. 

I look forward to your feedback!

Presently, anyone with any level of experience or training can enter data and edit existing data in the FamilySearch.org Family Tree. For years now, some of us have been advocating that there be some minimal training requirements before any changes can be made to the Family Tree. This training could cover such issues as entering data, changing existing data, merging duplicates, adding GEDCOM files, and other pertinent topics. The training already exists but it is not presently noticeably available or required before taking part in the Family Tree. This is not viewed as a way to prevent people from using the Family Tree, but to help upgrade the content and help prevent wholesale unsupported changes. 

You can comment on this post or contact me directly if you have any questions.