It is not as if I were trying to learn the answer to a question as profound as the one posed in Douglas Adams’ sci-fi classic “A Hitchhiker’s Guide to the Galaxy”, but I would say it was close in terms of profundity. The series of questions I recently asked of two AI chatbots involved a relatively large number compared to the famous 2-digit response given by the supercomputer, “Deep Thought.” The answers I received were not as cryptic as those from the fictional computer built by hyperintelligent pan-dimensional beings. However, what I got back from one AI chatbot was not only wrong, but it was, to reference another sci-fi classic, “highly illogical.”

Recently, I was trying to understand the theory of pedigree collapse, about which I had previously written. Pedigree collapse is a genealogy concept that occurs when two individuals with shared ancestry reproduce, resulting in fewer distinct ancestors in their offspring’s family tree. This commonly happens when an ancestor’s parents are cousins. For instance, if two first cousins have children, their offspring would have a maximum of six great-grandparents instead of the expected eight, resulting in a reduction of ancestors, known as pedigree collapse.

Ostensibly, a family tree is a binary tree wherein the number of ancestors in each later generation is obtained by raising 2 to the power of n, where n is the number of generations. The formula for determining how many ancestors we have in each generation is x=2^n. For example, to calculate the number of ancestors that we would expect to be in our great-great-great-grandparents’ generation, we would raise two to the power of five: 2^5=32.

Last year I produced a video showing how I am related to Charlemagne. He was a 37th great-grandfather, which means we are separated by 39 generations. (TIP: to quickly count the number of generations, sum the number of greats and then add 2; one for your parents and one for your grandparents.) Suppose my family tree is a strictly binary tree in which the individuals in each generation are distinct. In that case, I expect to find that 39 generations ago, 2^39 of my ancestors were living on planet Earth. That number, standing for the number of my ancestors 39 generations ago, is not “42,” but rather it is 1, followed by 12 zeros – one trillion.

The problem is that the population of the Earth in 800 AD – the year that Charlemagne was crowned emperor of the Romans – has been estimated to be between 200 and 300 million people. This number is based on various factors, including historical records, archaeological evidence, and population models. At that same time, the population of Western Europe, where, according to DNA, all my ancestors were living, is estimated to have been between only 15 and 20 million people.

While pondering this paradox, I posed the same questions to both the Bing AI chatbot available via the Microsoft Edge browser and to Google’s AI chatbot offering, which is called “Bard.” See Bing or Bard.

The responses that I got back from the Bing chatbot were quite helpful. To each question, I received a summary response, plus several links to sources. This included a PDF file of a paper published over 20 years ago that mathematically demonstrated how all Americans of European descent must be related to Charlemagne. Also, there was an article from 2013 about how researchers in California conducted a genetic survey involving the DNA of persons of European descent and found they all share a most common recent ancestor (MCRA) going back only 1000 years (see All Europeans are related if you go back just 1,000 years, scientists say).

I asked the Bing AI chatbot, “If there were only 30 million people in Europe in 800 AD, are all of them my ancestors?“

Bing responded: “No, not all people in Europe in 800 AD are your ancestors. However, it is possible that you are related to all Europeans who were alive at that time. According to a study by researchers at the University of California, San Diego, anyone alive 1,000 years ago who left any descendants will be an ancestor of every European.” One of the sources to that response was a blog post written by a professor of biochemistry at the University of Toronto titled “Are you a descendant of Charlemagne?“

One question I posed to both chatbots was: “I calculate that there are 39 generations separating me and Charlemagne. Going back 39 generations, I calculate that I would have 2^39 ancestors; in other words, 1 trillion people. That is more than the number of people who were living in Europe in the year 800 AD. How can I explain this paradox?“

The answer I got from Bard was not only wrong but perhaps a little condescending. In addition, part of the response made no sense whatsoever. Here is what Bard told me. I have highlighted the part that defies logic. Also, note Bard does not supply any references or sources for its responses.

Bard responded: “There are a few possible explanations for this paradox. One possibility is that you are overestimating the number of generations that separate you from Charlemagne. It is possible that some of your ancestors were related to each other, which would reduce the number of generations. Another possibility is that you are overestimating the number of ancestors you have. It is possible that some of your ancestors were not fertile, or that their children died before they had children of their own. Finally, it is also possible that there were more people living in Europe in the year 800 AD than we currently estimate. This is a complex issue, and there is no easy answer.”

I do not want to criticize AI chatbots too harshly as they are a recent technological advancement. As humans, we often face challenges when encountering new and unfamiliar things. So, in the spirit of cooperation and since it was offered, I asked Bard to help me finish writing this article by requesting that it provide me a summary of how AI chatbots can be used for researching family history.

Here is what Bard told me:

AI chatbots can be used for researching family history in a number of ways. For example, they can be used to:

• Access and analyze large amounts of data. AI chatbots can be used to access and analyze large amounts of data, such as census records, birth and death records, and newspaper articles. This can help genealogists to find information that would be difficult or impossible to find on their own.
  
• Ask questions and get answers. AI chatbots can be used to ask questions and get answers from experts. This can be helpful for genealogists who are stuck on a particular question or who need help interpreting a document.
  
• Translate languages. AI chatbots can be used to translate languages. This can be helpful for genealogists who are researching families from other countries.
  
• Connect with other genealogists. AI chatbots can be used to connect with other genealogists. This can be helpful for genealogists who are looking for help with their research or who want to share their findings with others.
  

However, there are also some concerns about the use of AI chatbots for researching family history. These concerns include:

• Accuracy. AI chatbots are not always accurate. They can sometimes make mistakes, such as providing incorrect information or misinterpreting a document.
  
• Bias. AI chatbots can be biased. This means that they may be more likely to provide certain types of information over others. For example, an AI chatbot that was trained on a dataset of historical records from the United States may be more likely to provide information about white Americans than about black Americans.
  
• Privacy. AI chatbots can collect a lot of data about the people who use them. This data could be used to track people’s movements or to target them with advertising.
  

Overall, AI chatbots can be a useful tool for researching family history. However, it is important to be aware of the limitations of AI chatbots and to use them with caution.