What to make of AI in family history research

After researching both sides of my family for several years, I don’t mind admitting I have a few brick walls that, despite diving into deep rabbit holes and subscribing to various sites, remain firm and impenetrable. I was having a chat with my son about this, to which he replied, ‘Have you tried using AI?’ The look of horror on my face precluded further response, to which he laughed.

‘Mum, I’m not suggesting you use AI to write your stories, but it might save you hours of research. What do you want to know? Let’s try one problem and see what happens.’

Okay, I thought, let’s give it a go. Was I nervous about it? Very much so. Am I still concerned about dipping into AI? Very much so.

Recently I had started delving into a male convict who was transported to Van Diemen’s Land in 1820. Thanks to another descendant who has put in decades of diligent research into our shared Tasmanian ancestors, I learned that he was Irish. Because this ancestor was arrested in Devon for highway robbery in 1819, I assumed he was from Devon. I had never seen his place of birth on any of the transportation records I had accessed, but it was there on one document shared by my fellow family historian. I am always ‘time poor’ and tend to dip in and out of my ancestors’ lives periodically. Does this sound familiar?

Another notable fact in that document was his occupation as a soldier in the 57^th Regiment. Wow! So our Irishman was a soldier in the British Army but somehow ended up in Devon, committing crimes which led to his demise. I’m still not completely sure how and when this soldier ended up in Devonshire. Naturally, I was intrigued to learn more about the 57^th Regiment.

The document giving his place of birth and occupation told me that our ancestor was an enlisted man in his early twenties when he was convicted for highway robbery. He was sentenced to death, but luckily, the sentence was commuted to life and transportation. I shared this new information with another descendant/family historian, who promptly reached out to a friend who has subscriptions to Find My Past and the British Newspaper Archive. From that, we learned our recalcitrant soldier was in uniform, with an accomplice as they assaulted a chap for his waist coat.

When writing about ancestors, I’m always mindful of context, so I wanted to learn more about the 57^th Regiment of Foot. This was the focus of my AI search, which produced a ‘dossier’ on the regiment that for me was really encouraging.

The AI platform my son uses is Claude, which he uses solely for research purposes. After inputting what I knew and my question/s about the regiment and the soldier, Claude produced a detailed report in four parts, along with a summary of what is proven and what is not:

• Part One was all about the 57^th Regiment, structuring details around: origins and early service, demography of the regiment, and the various battles that the regiment engaged in.
• Part Two focussed on our soldier/convict, a timeline of key events including some very interesting background on the convict ship Coromandel and the connections missing or otherwise between the convict and the ship and regiment.
• Part Three was an overview of how soldiers came to be transported; as in court-martials, convictions in civil courts, and the post-war surge – context of 1819.
• Part Four was a two-column table of record/series and what it would show, and details of where to source information. Part Four also provided practical search tips for both free and subscription-based sites, as well as guidance on how to word a search query effectively.
• Concluding summary was another two-column table; one with data confirmed by record and the second with what was unverified (referring to my query that did not include sources.

I have to say I was excited whilst at the same time equally concerned. Not for a minute do I assume that everything produced in the report is unbiased and completely accurate. If I were to use another AI provider, the data might be similar, or then again, perhaps not. My AI journey is in its infancy and, like everything else on my plate, will be sporadic.

Does the report give me a starting point for further inquiry? I would say ‘yes’, as producing the information would undoubtedly have taken me many hours, if not days. And while I can only commit to snatches of time, I am willing to use these reports as tools for useful guideposts. Under no circumstances, however, would I entertain copying and pasting any of the text into my own writing. That use of generative AI gives me the horrors. I think that is what scares us writers and researchers witless.

What cannot be understated, though, is the collegial sharing between like-minded family historians. That support will remain invaluable and is a celebration of human endeavour and critical thinking. Feedback from fellow historians has forced me to critically examine some of my findings, in which I most likely have jumped to a conclusion because certain dates lined up.

But as a tool for research triggers—in my very limited experience — I think AI earns a place in our toolbox. It just might help me dive into better rabbit holes where I still have to put in the research work. As with mobile phones, the internet, and computer tools such as spell check, people will increasingly come to rely on this technology, but I fervently believe it should never replace human creativity, endeavour, and critical thinking.