Stretching the truth: AI and ancestor overviews
Our online search habits are changing rapidly, and the days of trawling through pages of Google searches are seemingly over. Instead, whenever we type something into Google, we are immediately met with an AI generative overview of whatever it is we’re looking for. Need to know how much water to add to your rice? An AI overview will instantly tell you the ratio. Looking for holiday inspiration, and you want to know where’s warm in January? An AI summary will provide a list of suggestions, all accompanied by a breakdown of flight times from the UK and average temperature for the time of year.
AI overviews are even creeping into the world of genealogy, with summaries of ancestors readily generated—you know the ones, the brief outline of an individual’s life with the key dates, sometimes accompanied by a generic image or silhouette. I am often impressed—and bemused—at how they make such little information go so far! This is particularly the case with the ones produced by Google, where a summary is often the first thing you see whenever you search for an individual, the caveat being that there must be enough info available online about them for an overview to appear. Sometimes these can be helpful—such as when you’ve overlooked a potential source—but they also have a habit of muddying the water.
This was the case when researching Henry Seatle, the nineteenth century curate of Blawith, Lancashire. Ancestors with clerical roots often leave a paper-trail ––more so than your average ancestor—and Google can be a great way to find them. I wasn’t sure if anything would come up for Seatle as he was only the curate of a small, rural township with a fairly modest living. To my surprise, though, an AI overview popped up with a summary of his life—date of birth, children, and time served as curate. Most, if not all, could be gleaned from the records, but one aspect did stick out, and that was his “famous historical testament” on the working conditions of “pauper apprentices” in a local mill in 1816.
As you can imagine, I was instantly intrigued, and I couldn’t wait to track down the source. Yet when I actually found the source, the reality was rather underwhelming. It was, in essence, a rather generic statement made by Seatle confirming that the apprentices attended the parish church on Sundays—not quite the “notable artefact” in the historiography of child labour promised by AI. Instead, it was a formulaic statement not too dissimilar to the ones sent by incumbents up and down the country reassuring their archbishop that all was well in their parishes—hardly a groundbreaking insight on the health and welfare of pauper apprentices!
It is also worth noting that when I did the same Google search a week later, a different AI generated overview appeared—one which didn’t reference his testimony. Then when I googled him again two weeks later, the summary had changed once more—this time framing Seatle as an “influential English clergyman”. Whilst I’m sure Seatle probably had achievements which didn’t make it into the records, it would be stretching the truth to suggest he was influential based on the single source referenced!
This is not to say the AI summary wasn’t useful—after all, it did point me in the direction of the source in question. Yet it was another reminder to avoid taking AI overviews at face value, mainly because they have a habit of exaggerating or overstating the significance of the evidence. It is also worth playing around with keywords in Google to see the various shapes that the AI overviews take, and also doing this regularly as AI develops and new sources are added. A good rule of thumb, though, is to always seek out the original and verify what AI is telling you—you may well find yourself reaching wildly different conclusions.