Testing MyHeritage GAIA: Can AI Help With My Genealogy Brick Wall?

On October 21st, I received an email from MyHeritage introducing GAIA. GAIA is an AI-powered research agent. The email included the following:

"GAIA plans and executes family history research tasks with the accuracy and thoroughness of a professional genealogist. By asking targeted questions, searching records across MyHeritage and external databases, and cross-referencing every detail, GAIA breaks through your toughest brick walls and delivers high-confidence leads."

The email indicated I could engage in a limited number of conversations free to check it out. They've also introduced a new OmniAI subscription plan, with expanded use of GAIA, but I've been unable to find a cost for this plan. Based on James Tanner's post Introducing GAIA -- A Real AI Research Agent from MyHeritage on Genealogy's Star, it appears some existing plans do include limited access to GAIA. After my one sample test described below, the usage indicator showed 61% of my GAIA allowance used. I have the existing Omni plan.

All my internet searches return the standard existing Omni plan pricing. Personally, I'm not prepared to upgrade anyway. MyHeritage is a resource I use occasionally, not frequently. I already struggle to justify the cost of the Omni plan, given my sporadic use of the site.

Although dubious, I headed over to try it out. I've tried other AI-based genealogy searches with ChatGPT, and while the results might be useful for a new family historian, and they may have provided solid research guidance, they weren't terribly enlightening for my personal brick walls. As you likely know, I've been searching for Abijah LeRoy for more than a decade. It was a great way to try GAIA, so I prompted with:

"I have been looking for Abijah LeRoy (possibly LeCroy) who lived in Georgia. He was married to Nancy (Williams) LeRoy, and had several children born between 1839 and 1847. I have found census records for his wife and children, but nothing for him at all. Abijah is the name noted on his son John Robert LeRoy's death certificate in 1927, but it may be a nickname of some sort. I have seen records for a Thomas LeCroy but nothing sufficient to be sure it's the same person. He may have lived in Habersham or Whitfield Georgia. I also found a lawsuit where Nancy claimed a Thomas LeCroy owed her money, but it would be odd for it to be her husband. Looking for anything to help properly identify the name of her husband and find more about him."

While GAIA was happy to confirm details I already knew, and to agree with my reasoning on names I didn't find viable, it didn't produce any significant new information. It provided most of the same detail I already knew, and resources I've already checked. I did give it a quick go checking for probate records in Habersham County, Georgia, since it suggested it might be a next step—but it found nothing. I made one last attempt by following the suggestion to look at the 1840 and 1850 censuses (although I already know 1850 does not show him with his family). GAIA found other census records for Nancy (information I already have) but didn't find anything on Abijah. It did provide seemingly useful next steps—if I hadn't already done those available to me online. Certainly there may be more offline research I could engage in, but AI or no AI, I'm likely not going to do it anytime soon. 

I do want to point out, if records aren't available online, an AI agent searching online won't be able to retrieve them on its own. The absence of new information does not necessarily indicate GAIA (or any AI model) didn't do a thorough search and review. None of us can magically make missing or inaccessible records appear. We can follow the best-laid research plan and still come up with nothing. There are so many reasons a person may not appear in records. As frustrating as this may be, not every person will be found, and not every detail will be known.

For my personal, very brief test of GAIA, I think it could be useful for those who prefer more personalized direction, but for me, it's not returning compelling enough results. So, this is where I stopped my test. I think it did a fine job of suggesting research avenues, even though it didn't produce a new lead. The issue is—not all research dilemmas can be overcome by searching online. On the positive side, the interaction felt "friendly" (or as friendly as a computer to human conversation can be). It did provide solid options for research suggestions, and it did try to do the research if I asked it to. For some users, I suspect it would be a nice complement to traditional research. It simply isn't for me.

One benefit may be the ability to search within MyHeritage records. General-purpose AI tools may not have direct access to subscription genealogy records. Many genealogy records are behind paywalls, limiting the AI responses. GAIA may break through part of this barrier for MyHeritage users. MyHeritage has long been at the forefront with AI photo editing tools. And they don't seem to be slowing down with the introduction of GAIA for AI research. Other sites will most likely follow suit, but when and how will depend on their own agendas. I truly believe AI has many uses within genealogy, and as it evolves it may very well become a staple product for family historians. But until far more record sets are made readily available, those "A ha!!" moments will likely still come primarily from personal research. 

Image by Gerd Altmann from Pixabay

Disclosure: I used ChatGPT to assist with a title and search description, and to review this post for clarity, punctuation and grammar. Unless otherwise indicated, the content and all opinions are my own.