What Is All This Nonsense?

Lately, I have been reading a lot of nonsense. I’m not talking about social media posts or news stories. Rather, I keep running across information about our Hagenbuch family that makes no sense to me.

For example, here is some text found on a newly-created Findagrave.com memorial for our ancestor, Andreas Hagenbuch (b. 1715):

Source: Article got to the point!!!!!!!!!
Rethinking Andreas children. Published March 6, 2018. Updated March 6, 2018
The order of the children according to the Last Will and Testament found on Andreas Hagenbuch
Note: That goes for the U.S. Census records found on Andreas Hagenbuch’s descendants. The records are based off of assumptions. The order of the children listed on some U.S. Census records is not the actual order of the children who were born. So the U.S. Census record becomes faulty. You really have to analyze these records to determine the vital records of the family of Andreas Hagenbuch, his wives and children. On the church records because these are actual names you are looking at totally different families. So these children are linked to the wrong families. On the church records it is assumed it is due to a misspelling. Not the real person. So how are you to analyze these church records? Another family or a misspelling? This has been done many times on the church records. The sponsorship of the children was based off of rules set in place by the Lutheran Church. These rules had to be followed by the parents. If you look up the definition and actually think about it in the eyes of the parents. What do you really see? I took my religion very seriously. I was required to practice it. Did I really practice it? Or did I blow it off because that’s how things were done. I went with everybody else. So did everybody else.

After reading the above, I was left scratching my head. No only does it say very little about Andreas or his children, but it also appears to reference a 2018 article found on Hagenbuch.org which is cited as “Source: Article got to the point!!!!!!!!!” Who had written this memorial and why?

Portion of the Find a Grave memorial for Andreas Hagenbuch that was created by an anonymous user

The first question has an immediate—if unfulfilling—answer. The new memorial for Andreas was created by anonymous user 51019174. This Find a Grave user account is three years old and has created over 200 memorials on the site. No personal details are provided by user 51019174 in their profile, leading me to question if they are even human.

For comparison, here is some text from the Find a Grave memorial for Henry Hagenbuch (b. 1737):

[Henry] was commissioned a lieutenant on May 22, 1775 of the Salisbury Company, with a total rank and file of one hundred men. On July 9, 1776, he was commissioned captain of a company, which was part of the Flying Camp of 10,000 men commanded by Colonel Hart. On Aug. 6, 1776, Captain Hagenbuch and company were encamped at Perth Amboy, New Jersey, and on August 27th, of the same year, they participated in the battle of Long Island, in which some of Captain Hagenbuch’s men were wounded. He continued to be active with the associators and militia of Northampton county as late as April 23, 1783.
Hagenbuch family history & genealogy. Submitted by Peggy Spengler-Moser.

Henry’s memorial was created by Amanda (Bowser) Ellinger, who is Find a Grave user 47378067. The memorial contains specific details about Henry’s life and mentions a verifiable source, Peggy Spengler-Moser, who is a genealogist I met while exploring Christian Hagenbuch’s (b. 1747) property in Northampton County, Pennsylvania.

I was starting to believe that anonymous user 51019174 might be an AI-powered bot. The bot seemed to be scraping content from Hagenbuch.org and converting it into a generic summaries that lack relevant details.

AI-generated content seems to be everywhere. Estimates are that as many as 20% of Google’s search results contain text and images that were created by AI software. Unfortunately, AI-produced content isn’t always of the highest quality, particularly when relates to a niche subject matter, like a specific family’s genealogy.

A portion of a conversation with ChatGPT about Andreas Hagenbuch (b. 1715)

I wanted to try out just how good AI-generated information can be, so I signed into one of the best AI tools—ChatGPT—and asked it a basic question: “When did Andreas Hagenbuch come to America?” I received the following response:

Andreas Hagenbuch immigrated to America in 1737. He arrived in Philadelphia, Pennsylvania, on September 26, 1737, aboard the ship St. Andrew Galley.

Unfortunately, two of the most important facts in this response are incorrect. Andreas Hagenbuch landed in Philadelphia, PA on September 18, 1737 and had sailed aboard the Charming Nancy. These facts come directly from the primary sources mentioned in this article. ChatGPT is confusing basic details about the Charming Nancy with another ship, the St. Andrew Galley. Both ships arrived in Philadelphia around the same time and were captained by brothers: John Stedman on the St. Andrew Galley and Charles Stedman on the Charming Nancy. However, we definitively know that Andreas Hagenbuch and his wife, Maria Magdalena (Schmutz), disembarked on the latter of the two.

I have heard that AI has the ability to sound authoritative on all subjects, even when it is unsure or incorrect. For genealogists this is especially concerning. Even before the development of AI, misinformation in genealogy was a problem. My father, Mark, and I have spent years researching, debunking, and revising long-held fallacies about our Hagenbuch family.

Before logging out of ChatGPT, I wanted to conduct one final test using its DALL-E imaging functionality. I began by asking it to find a picture of Andreas Hagenbuch. When it informed me that there were no known portraits of him (which is true), it suggested I ask it to create one. I did so and received the following image:

Andreas Hagenbuch (b. 1715) as imagined by ChatGPT’s DALL-E feature

Honestly, I laughed at how much this Andreas looked like a fashion model version of Thomas Jefferson! I had one final test in mind, and I asked ChatGPT to find a picture of Hiram Hagenbuch (b. 1847). Again, it told me that there was no known picture of Hiram, which is incorrect. Below is the best photograph of Hiram in the Hagenbuch Archives:

Hiram Hagenbuch, 1872

Like before, ChatGPT suggested that I ask it to create a picture of Hiram Hagenbuch, which I did. I received another laughable result. Can you see the resemblance to the real Hiram?

Hiram Hagenbuch (b. 1847) as imagined by ChatGPT’s DALL-E feature

I understand that ChatGPT, DALL-E, and other AI technologies can excel at certain tasks, such as expediting the creation of computer programs and identifying tumors in medical scans. Yet, when asked to answer basic questions about our family’s history, these technologies fall flat. Simply put: they shouldn’t be used for this purpose.

This brings me back to the second part of my original question. Why does Find a Grave have AI-generated content on memorials? One theory is that its parent company, Ancestry.com, is experimenting with AI tools that automatically create and update memorials. While possible, this seems unlikely. The quality of the content added to the new memorial for Andreas Hagenbuch is quite poor. Additionally, one would hope that Ancestry would be transparent when updating the site and avoid creating anonymous user accounts.

Another theory is that an unknown, third party is posting the content and using an AI-tool to automate its creation. This seems more likely. Why exactly it is doing this is up for debate. Since Ancestry acquired Find a Grave in 2013, it has added game-like features to the online database of gravesites. These features reward users for adding memorials, photographs, and other information. It is possible that this third party is experimenting with software that automates the creation of memorials and content for these.

Genealogical research is hard, time-consuming work. It relies upon reliable source material and requires careful attention to detail. Misinformation has always been a problem in genealogy, and family history books from over a century ago often contain inaccuracies. Nevertheless, the ubiquity of modern AI tools poses a unique threat—one that could turn everything we have learned about our Hagenbuch relatives into nonsense.