Let’s Try Some Artificial Intelligence On My Genealogy

Note that I have tried to show ChatGPT’s written results so that they appear separate from my own. I’ve done this by either putting it in bold or if I couldn’t in WordPress I’ve used a separator.

I’ve been playing around with the paid version of ChatGPT, and today, I thought I would show you how it does with some genealogical records. Below is a photo of the Beaton family that I scanned from the book Lilacs and Limestone by William John Patterson.

Beaton Family- from Lilacs and Limestone, written by William John Patterson

I decided to upload the photo to ChatGPT with the following prompt. (I’m still learning as I go, and I know the prompts are important, so these could be better).

Look at the attached photo and provide the approximate date and style of clothing and any other genealogical information that could be obtained from the photo.

If you didn’t watch the video, this is what it said:

Examining the photograph you’ve provided, there are several elements that can help determine an approximate date and provide context from a genealogical perspective:

  1. Clothing Styles: The women in the photograph are wearing dresses with tight bodices, high necks, and long, draping skirts, which is characteristic of the late Victorian era. The men are in sack suits, which also align with this period. The presence of bustles on the women’s dresses and the style of the men’s suits suggest a date in the 1880s to the 1890s.
  2. Hairstyles and Accessories: The women’s hair is styled away from their faces and is typically swept up, which was the fashion of the time. The children’s clothing also fits with the late Victorian trends.
  3. Photography Style: The posed group in a studio with a painted backdrop and props (like the chair and curtain) indicates a formal portrait session, common in the late 19th century.
  4. Children’s Ages: By estimating the ages of the children and considering the adult’s apparent ages, one might infer potential birth years and thus, approximate years for the photograph.
  5. Family Structure: The photo seems to be of a single family, which can give clues about family size and relationships during that time.
  6. Other Details: The small details like jewelry, the chair design, the carpet, and the drape patterns can sometimes be traced back to specific time periods or geographic locations.

For genealogical purposes, you could use this information to search for fashion trends, census records, and other historical documents from the 1880s-1890s to find more about the family. It’s also worth noting that family portraits of this kind were often taken for significant events, which could be an avenue for further research, such as weddings, anniversaries, or reunions.

Please remember that while these clues can provide a likely time frame, without additional context or records, it’s not possible to give a precise date. If this photo is part of your family collection, it might be worthwhile to compare it with documented dates of events and known ages of the individuals in the picture to narrow down the date further.

I actually tried this a couple of times and each time I got a slightly different answer. There were a few points that weren’t mentioned in the above.

The family appears to be well-dressed and probably middle to upper-class, given the quality and style of their clothing.

This type of photo would have been used to document the family at a particular time and would have been shared with other family members or kept as a keepsake.

The photo may have been taken in America, Britain, or another English-speaking country, given the fashion styles that were heavily influenced by English trends of the period.

Now, let’s look at what artificial intelligence can do with handwriting. I’ve been told that handwriting isn’t quite there yet, but we’ll give it a go. In this case, I’ve used a screengrab of a will. Keep in mind that Ancestry used artificial intelligence to transcribe the 1950 US Census and the 1931 Canadian Census, and they did so at record speed.

Here’s what ChatGPT came up with.. and although it had some problems, I might be able to handle some of the rest. Note that it also offered some thought about the Hudson Bay Company.

Codici[illegible] [Illegible] me it known unto all men by these presents that I Robert Goodwin stand at Albany fort Johnson esq in the service of the Hudsons bay company London having made and declared my last Will and Testament in Writing and appointed in the hands of John [illegible] [surname] Gentleman residing at [illegible] street London I now do ratify and confirm all and singular his my companions in this codicil and do nominate [illegible] said last will and so give and bequeath to the apron? John David Towne John Barum? Gentleman of [illegible] in Suffolk and Robert Furnis? Gentleman of [illegible] in Suffolk the Executors under my said will the sum of three hundred pounds sterling piece current money in trust for the following purposes that is to say the interest thereof going fifteen pounds per annum unto her Sarah [illegible] Brookes part of company London to be equally divided between Caroline Sarah my and Peggy my natural daughters

The transcription above includes some uncertainties due to the illegibility of some portions of the text. The handwriting is indicative of a style commonly used in historical documents, and while some words are clear, others are obscured or difficult to decipher without context. Documents like this can offer a wealth of information, including names, locations, and relationships, which are all critical pieces in genealogical puzzles.

It’s also worth noting that the monetary amounts and the mention of the Hudson’s Bay Company suggest economic status and perhaps a connection to trade or colonial activities, which can be a significant point of interest in your genealogical narrative.

I also tried to have it translate a photo of a French document. This was a failure, although it did try two times and then gave me some advice.

The OCR process has been interrupted due to a timeout, which indicates that the process took too long to complete. Unfortunately, I’m unable to transcribe and translate the document using the methods available to me at the moment.

However, given your expertise in genealogy, you might consider using dedicated OCR software that specializes in recognizing older types of print and handwritten texts, which might be more efficient for this type of task.

It made me laugh when it said that, given that I’m an expert genealogist, I should consider using OCR software, which I’ve done in the past with this same document. In this case I upload the document to OneNote. This is a page for La Famille Janelle.

With a quick right click, I can choose “Copy text from the picture,” and then I paste it to the page. I get the following. This is a screen grab, but it’s text that I can then put into Google Translate.

I decided to give ChatGPT one last chance at this document, so I asked it to translate the French text into English. Here’s what it told me, including some grammatical thoughts.

The translation of the French text to English is as follows:

Mr. J.-Émile Janelle,
My dear Cousin,
You ask me for a preface, but it is rather a thank you
that I want to write to you, for you have accepted the task of
realizing, thanks to patient studies and tireless research,
the beautiful dream that my filial piety once cherished.
I believe I owe you a gratitude that measures up to the happiness
I felt upon reading your work.
You, who keep, like us, in your heart the cult of
memory, I know well that your first reward, you
found it in your labor itself: from these yellowed sheets
piously preserved in the family papers, there rose towards you
like a Perfume of virtues that embalmed
your soul, while at the same time there escaped a discreet call
to a life of labor, of faith, of valor, and of honor,
which was the life of those who left us, with their name,
the legacy of a blessed memory. But it is for me a
need of the heart to thank you again, at the moment when
you are about to deliver the impression of these lines written
with a filial, sincere, and careful pen to snatch from the most distant
archives the slightest particle of historical truth, even
at the price of countless journeys, steps, and vigils.

Please note, the French text includes some typographical peculiarities (e.g., “Premi?re” which should be “première”) and “voUs” which should be “vous”, and I’ve translated assuming these to be typographical errors.

So, what does this all mean to us as genealogists? For me, it provides context and food for thought in a photograph. Perhaps noticing something I hadn’t though of. Eventually, it will be able to transcribe documents even better, but even as it is, I can work with what I’ve been given already.

What artificial intelligence is doing is it’s saving me time, and if I have more time to do genealogy then I’m all for that.

If you have enjoyed this or other blogs that I’ve written, then I hope you’ll join my pack. By joining, you’ll not only get updates on blogs, but you’ll also receive special offers from my genealogy friends, and they’ll all come to your inbox.

Comments are closed.