AI and Ancestry Hints

While AI and OCR recognition have been used for quite some time, it's only been in recent months I've seen an increase in inaccurate hints on Ancestry. It seems to correspond with the widespread increase in the use of AI across the board. While I absolutely appreciate more possible records for my relatives and ancestors, you never know when an "ah ha!" hint will appear, I can't help but notice how many new hints are not relevant. AI considerably reduces the necessity for human intervention and indexing. Combined with the magical algorithms, one would certainly expect many new results as this technology takes hold. But it's far from perfect. 

In the past, I'd estimate 95% of the hinting was pretty accurate for my tree (not including user uploaded media potentially appearing more than once). There were always a few scattered incorrect records suggested, but by in large, I was quite happy with results I received. I suspect many of those incorrect hints were the result of the algorithm picking up on records attached in other trees. Now, I'm finding more inaccurate records than ever before, especially in the hints pulled from newspapers and yearbooks. It seems like the combination of AI and algorithms may not be filtering as well as I've been used to. Now, I have a "reject rate" of closer to 15%.

I've found numerous newspaper hints where the person in my tree would have been a youngster, far tooo young to be the subject of a newspaper article. And many appear to be picking up on a spouse (showing a Mrs. [fill in the blank]) when the person being researched is not female (and most often has no tie to the article referenced). I've also noticed the algorithm seems to combine names (example: Mr. Bill Jones went to the home of Mr. Wayne Smith might produce a hint for Bill Smith in my tree). When it comes to yearbooks, the algorithms are often pulling women's married surname instead of their maiden surname (and while it's possible these women might have been married while still in school, it's unlikely.)

I'm a big fan of AI overall and I use it daily as my "assistant", helping me format, translate/transcribe and extract data. I absolutely believe it has a very valid use when it comes to scanning large data sets to provide us with new hints. I just wish they'd tighten up some of the parameters, or perhaps run a second scan of the potential records to weed out some of the inaccuracies. I fear click-happy new researchers (I was one of them back when I started!) will attach anything and everything without proper vetting, causing inaccuracies to populate even faster.

Of course, it is (and always has been ) the responsibility of each researcher to determine if records apply or not. This is not new! Only you can determine if you believe a hint is accurate or not. Over time, I suspect we'll see the inaccurate results decrease as people reject irrelevant hints, and algoritms are tweaked, I believe we'll ultimately see overall improvements in hinting. But for the short term, we may need to be a little more careful evaluating new hints. 

One final thought...it's important to note, correlation is not necessarily an indication of causation. I could be totally wrong in my observations. Perhaps it's simply the addition of new newspapers and yearbooks. Perhaps I've been underestimating my previous "reject rate". Have you seen any trends in Ancestry hinting since AI has become more widespread? I'd love to hear your comments and thoughts.

Image by Gerd Altmann from Pixabay

Testing Ancestry's Photo Facial Recognition

One of the newer Ancestry.com updates is AI assisted facial recognition. I saw it mentioned in an email and quickly clicked on the button to test it out! I was directed to my media gallery, but was presented with no information on actually using the feature. A quick hop over to the help pages didn't yield any clues, so I started playing around.

From my media gallery, I selected a photo where I knew who was shown. In this picture is my mother as a young girl. To start facial recognition, first click on the "tag" button (shown with a red arrow in the photo).

Then click on the face of the person you want to search. A box will appear around the face and a pop up will appear where you can select someone in your tree, or choose "Recognize Ancestors". I chose "Recognize Ancestors". CAUTION: If you select someone in your tree, it will simply tag the selected person in the photo with the name chosen. It will NOT compare the photo to other photos of the person you may already have in your media gallery.

You'll be presented with possible matches for the person you selected, and an analysis of how viable the match suggestion may be. AI is searching other media in the database to look for potential matches. In this example, the top option is labelled "Great" for this match (but it does show it's a match to my own tree), none the less, it did match to the correct person, my mother, who was already in my tree. It matched her face to exactly the same photo, so it was easy to confirm AI did a good job analyzing this photo. 

While the first photo worked quite well, when I tried it on another photo of my mother, you can see from the results below, it did not do so well at all. I was presented with two potential matches, but none of the options were listed as "Great", only "Good" and "Fair". And none of the suggestions were actually the person in the photo. 

I tried the same test with 2 photos on another person in my tree, and I had the same results. It was 50/50 on how well the AI did. To use this feature, the media item must already be saved in your gallery. I tried this on a regular photo hint, but could not select to tag the person. It makes sense to only allow you to tag media you've already saved. But, on the flip side, it would be nice to be able to do facial recognition without actually tagging anyone at all, in advance of saving a photo hint. We all know, some photos are not of the person the original poster attributed them to, and it could potentially prevent (or at least limit) people from perpetuating incorrect information if they could verify before saving. To some extent, if you already have media saved, there's a reasonable chance you already know who's in the photo. But certainly, there are times when there are unknown people, or you take the details of a photo at face value, and would just like a way to see if the info may be correct.

The takeaway? While this is a great new tool to potentially help identify unknown people in a photo (or perhaps confirm your suspicions), it should be used with caution. Ideally, you'll be searching for people who you already have a photo of in your tree, where you'll be able to do your own comparison to see if you think the AI is correct. As with anything AI, it's best to double check everything. This is a big step in a new direction, but it's not a magic bullet.

Image by Tumisu from Pixabay
Photos of Margaret Irene LeRoy are from my media gallery

Exploring the Past with a Glimpse of the Future: DALL-E and Genealogy

Image created with AI DALL-E at ChatGPT

One of the features in the expanding world of AI is DALL-E (available with a ChatGPT Plus subscription of $20 per month). This is the part of AI capable of generating amazing images from words (and sometimes a little imagination). For this post I simply asked DALL-E to create an image for my genealogy blog. The image on the left was generated, and I have to admit, I really like it. AI provided me with this text to go along with the image "Here's the image created for your genealogy blog. It's designed to visually encapsulate the essence of genealogy, highlighting elements...represent[ing] the discovery and curation of family histories. This should serve well for your blog, inviting readers to explore the...content you've selected."

I was a little short on time for my post this evening, so I asked ChatGPT 4 to provide me with a bit more information on using DALL-E for genealogy. While many are jumping on the AI bandwagon, others are understandably leery of this new technology. Other than the final paragraph (and the removal of the word "that" in several places, because my high school writing teacher drilled into my head "whichery, whatery and thatery don't belong in writing, and to this day one of my final proofreads is always for these words) the text below is substantially unaltered and was provided by using the following prompt in ChatGPT 4 (Note: Upon request AI also provided me with the HTML code to create the shaded quote box displaying the text used for the prompt, saving me having to search how to code it manually):

Prompt:    

You are a genealogist learning about artificial intelligence. Provide a short blog post about creating images with DALL-E.

"As genealogists, we often find ourselves sifting through the sands of time, piecing together stories from the fragments left behind by our ancestors. Yet, despite our best efforts, there are moments and faces...remain[ing] shrouded in the mists of history, leaving us to wonder about the world our predecessors lived in. Enter DALL-E, an artificial intelligence marvel...[with] the potential to breathe life into our historical imaginations.

DALL-E, developed by OpenAI, is a cutting-edge AI program capable of generating images from textual descriptions. This tool doesn't just create art; it opens a window into worlds unseen and possibilities unexplored. Imagine being able to visualize a 19th-century marketplace as described in an ancestor's diary or the intricate details of a family heirloom...long since been lost to time. With DALL-E, these figments of our genealogical pursuits can be rendered into vivid images.

The application of DALL-E in genealogy extends beyond mere curiosity or the desire to fill in the visual gaps of our family trees. It represents a fusion of technology and history, allowing us to:

• Recreate Historical Scenes: By inputting detailed descriptions from historical records, letters, or diaries, genealogists can use DALL-E to generate visual representations of the lives and times of their ancestors, offering a new perspective on the information we've gathered.
  
  
• Visualize Ancestral Artifacts: From descriptions of clothing, jewelry, to household items, DALL-E can help visualize artifacts...once part of our ancestors' daily lives, providing a deeper understanding of their personal and cultural contexts.
  
  
• Enhance Engagement: For those sharing their genealogical findings with family members or the public, images created by DALL-E can make historical narratives more engaging and accessible, bridging the gap between past and present.

While the potential of DALL-E in genealogy is immense, it's crucial to approach its use with a critical eye. The images generated are interpretations, not historical records. They are based on the AI's training and the specificity of the input it receives. Thus, while they can add color to our narratives, they should not be considered factual representations. Furthermore, ethical considerations regarding the depiction of individuals and events must be carefully navigated to ensure...the use of DALL-E respects historical accuracy and sensitivity.

For genealogists, DALL-E offers a fascinating tool to enhance our exploration of the past. It allows us to visualize, engage with, and share our historical research in new and exciting ways. However, it's essential to use this technology as a complement to, not a replacement for, rigorous historical research and analysis. As we continue to integrate AI tools like DALL-E into our genealogical work, the boundary between the past and the present blurs, offering us a unique opportunity to connect with our ancestors' worlds in a manner previously unimaginable."

As you can see, AI did a nice job providing the information I requested and included sound advice on using DALL-E to create images for genealogy. As with any AI output, reviewing it for accuracy is extremely important. In this case, AI seems to have done a fairly nice job. I  believe it's appropriate to note when images and text have been generated using AI, and I will always strive to do so (unless I've substantially changed the text output). For this one post, I used AI to create an image, HTML code and provide relevant text. At this time, AI does not have the "training" or resources necessary to do reliable factual genealogical research and citations. It is not a research shortcut. For now, AI is merely a potentially beneficial tool in the genealogist's toolbox. Once you understand the limitations and strengths you can put it to work for you (should you choose to do so).

Image created with AI DALL-E at ChatGPT

Don't Waste Your Money on this AI for Family History Course

In general, I've been a big fan of the Family History Daily courses. They're designed for the casual or amateur genealogist and usually offer lots of practical and useful information. I purchased the Complete Access package a few years ago when I was restarting my tree. This package grants me free access to any new courses as they're added. I was really excited to see they added an "AI for Family History" course recently, and a few nights ago I settled in to work thru it. 

I'll start on a positive note. If you have no familiarity with AI and using these new tools, the information is presented simply and in a way most people can probably understand. Several AI tools are compared, highlighting the strengths and weaknesses of each. By using the same set of prompts for each tool, they clearly illustrate (with plenty of images) where different tools do well, and where they run into challenges. The course covers a basic introduction to AI, privacy and security concerns, tasks AI can and cannot do, and a presentation on creating simplistic prompts. It also illustrates how refining prompts assists in better responses. After showing examples from the various AI tools, they round it out with a few genealogy tasks AI is currently reasonably well suited for such as transcribing, translating, visualizing data and a couple others.

Now for the huge negative - the price. The course is regularly priced at $129, but could be less if they're running a special. I would NOT pay for this course under any circumstances! Certainly not at $129...and not at any reduced price either. It only took me a little over an hour to work thru the whole course (admittedly, this did not include any time spent practicing, as I already have a basic familiarity with AI). If you happen to have a Complete Access package (allowing free access to new course materials), it may be worth an hour or so if you haven't spent any time at all trying out AI. But, most of the information provided can be easily gleaned from reading posts written by those currently using AI as a genealogy tool. The comparison of various AI sites really didn't add much substance, though I did learn the names of a few tools I'll probably never bother to access. This is truly a bare bones introduction If you have any hands on time at all with AI, you're unlikely to gain much in the way of new insights.

Are you interested in an introductions to this new technology but don't know where to begin? I can definitely point you toward some free options to get you started! I recently came across AI Genealogy Insights. This is a blog dedicated to exploring AI as it relates to use in genealogy. While written at a slightly higher level than absolute beginner, it's more in depth and gives better insights into using AI alongside your traditional research. Best of all, it's completely FREE! There's an entire section dedicated to Use Cases where clear information is presented on creating and using extremely well structured prompts to achieve usable results. The prompts illustrated in this blog are head and shoulders above what's presented in the Family History Daily course. I'd urge you to spend an hour reading this blog over spending even a dime for the course. ChatGPT also offers the OpenAI Help Center. Here you have access to considerable information about this popular tool and how to best use it.

As you probably know from my other posts, I'm not at all opposed to paying for subscriptions, courses, or anything else to help further my knowledge and research. Some materials and tools are absolutely worth the investment (both time and money). This course is not one of them. You'd be far better off spending $20 a month for ChatGPT Plus for 6 months than paying for this course. AI is simply a tool. I don't mean to burst the bubble, but it's not going to replace traditional research any time soon. It can definitely assist with many mundane tasks such as transcribing and translating (though you still need to double check what it provides as it does make errors and sometimes fabricates information.) At this point in time, my personal suggestion if you're interested in using this technology is to read some of the posts on AI Genealogy Insights and go hands on with ChatGPT. You'll be surprised how quickly you can learn to use this new tech to help you in your genealogy journey.

Image by Gerd Altmann from Pixabay

Palaeography: Deciphering Old Handwriting

palaeography: The study of ancient writing and inscriptions; the science or art of deciphering and determining the date of ancient writings or systems of writing.

Deciphering old handwriting can be really challenging! I'm getting better (albeit slowly), but some of the mid-18th century and earlier really stumps me at times. Practice is definitely key. I've dabbled with using ChatGPT to transcribe portions of old documents...with mixed success. Sometimes it gives me a relatively good transcription and I'm like "Oh! That's what it says". Other times, it simply informs me old handwriting can be hard to read and I might need to investigate further and use other documents to compare writing samples. 

Recently, I saw a post where someone shared a cheat sheet of sorts. I've been using it the last week or so, and it's been helpful. It's been posted on multiple sites (with no original source noted). If you haven't seen it and would like to save a copy, just click the image and save it to your computer.

I've also come across couple of free online courses/tutorials. I'm planning to work through these courses in the coming months as a part of my "continuing genealogy education". I'm starting to get back to the late 1700's and early 1800's on my paternal line, and it's becoming very apparent I need to improve my skills in this area. 

The National Archives (UK) offers Palaeography: reading old handriting 1500-1800 A practical online tutorial with 10 lessons and tips for further practice.

CERES COPIA (University of Cambridge, England) has a course comprised of 28 lessons: English Handwriting 1500-1700: An Online Course.

If you're researching other countries and languages the challenge can be even greater. Not only do you need to know styles of writing, but the language at the time. I'd encourage  you to search out some resources before you jump in. The FamilySearch Wiki currently has over 9600 entries on handwriting, covering lots of countries. It also includes word lists you may require in your endeavors. Simply use the search box to narrow the results for the country or language you need. Thus far, I've been lucky enough not to encounter other languages, but given a portion of my paternal line is German, I don't think I'm going to dodge it forever. My DNA shows some other European countries and if I'm lucky enough to find where the ancestors I carry with me came from, I definitely want to be prepared.

While brushing up on skills is important, it's unlikely to solve all the challenges of reading old handwriting. The writer may have been in a hurry, ink may be faded and let's face it...some people just have chicken scratch! Only with time and practice will deciphering old writing become easier. AI may one day be sophisticated enough to transcribe consistently, but it's still no substitute for learning how to read documents yourself. Even if you are successful with an AI transcription, you still need to compare the AI result with the original and verify it's correct. As you continue your journey back in time, take time to learn to read old script (no matter what language it may be).

 (Links to referenced resources have been added to my Genealogy Toolbox page for easy reference.) 

Image by Petra from Pixabay

A Life Marked by Tragedy: Victor Michael Crow

I've intended to write about Victor Michael Crow (1918-1975) for some time. He was the husband of my 3rd cousin 1x removed. Not a close relationship, but I was touched at the multiple tragedies he suffered. I've also been meaning to test out ChatGPT since AI is here to stay and I need to learn more about it. So today, I've merged the two tasks and this is the result.

To be clear, I love to write! I have absolutely no intention of letting AI write "for me"...but I might let it "help" me from time to time. For this post I asked ChatGPT if it could write a bio from the facts I supplied. I simply copied/pasted the facts from the person profile in my Ancestry tree. I was surprised at how well it did, but I wasn't totally happy with the first run through, so I asked ChatGPT to try again. There were parts of each result I liked, so I've merged them and made a few tweaks (including adding details from newspaper articles not included in the facts I provided and correcting a few dates where AI didn't have enough information to correct a newspaper date to the actual date incidents happened). Overall, I was really happy with how well it did! While AI can't replicate my sometimes dry and hopefully witty humor, for a piece like this, it was helpful. Without further ado...

Victor Michael Crow was born on August 30, 1918, in Lostine, Wallowa, Oregon. He was the son of Michael and Jennie Crow and entered into this world on a bright morning at 9:45 am. At the age of 23, he explored his talents as an actor. Following this, he pursued a college education, completing four years before January 30, 1942, showcasing his commitment to knowledge and personal growth. The winds of World War II swept him into the military, where he served in the Army Air Corps from January 30, 1942, to December 21, 1945. Starting as a Private, he rose to the rank of Sergeant, displaying dedication and valor.

Residing in Wallowa, Oregon, Victor's life took an unexpected turn on November 18, 1955. Tragedy struck Victor's family when his first wife, Mary Lou Greenwood, and daughter, Jeni Le Crow, passed away due to accidental exposure to freezing temperatures. This incident surely left an indescribable void in Victor's heart. Sadly, his young son found the bodies of his mother and sister. While the cause of death is not known for sure, the coroner believed Mary Lou may have broken her back or neck while climbing a haystack to get feed for livestock. He theorized Jeni Le may have followed her mother out to the barnyard and ended up in the freezing cold alone. As his mother had told him she would be out for the afternoon, 11 year old Philip was not worried when he arrived home from school and no one was home. He ended up going to bed. When he woke the next morning and there was still no one around, he went out to look for them and found the bodies. With his father away for work, he drove a tractor 2 miles to town to report the incident. I can't begin to imagine how their deaths affected both Victor and Philip. 

In the midst of life's challenges, Victor found love again. On June 12, 1959, at the age of 40, he married Elda Mae Childers in Boise, Ada, Idaho. Their union took place at Orchard Avenue Baptist Church, and was witnessed by Ethel T Crow (possibly his sister-in-law, but I'm not 100% sure) and L Warner Crow (Victor's brother). The couple made their residence in Joseph, Wallowa, Oregon, building a life together. However, fate dealt another cruel hand and heartbreaking loss. On February 3, 1961, less than two years after their marriage, tragedy struck again as Elda Mae Childers lost her life in a terrible car accident. Their car crashed into a pickup truck stopped on the side of the highway. The impact threw the truck across a bridge abutment and the Crow's car spun out of control, throwing Victor from the vehicle into the middle of the road. There was a small fire in the front of the car, but the fire department was able to extinguish it. Elda Mae died instantly of head injuries. They had been on the way to a basketball game where one of her brothers was a referee and Victor's two children were already in attendance at the game.

Undeterred, Victor continued his journey and, on April 11, 1971, at the age of 52 in Lostine, Wallowa, Oregon, he entered into a third marriage. Victor joined his life with my relative, Ariss Jeanette Jones. The officiant for the ceremony was William B Watson, a clergyman. In the following years, Victor was actively involved in ranching, and he and Ariss made their residence in Enterprise, Wallowa, Oregon. Victor continued his ranching occupation until his passing. He filled his time as a member of Standley Masonic Lodge 113 AF & AM of Wallowa, Baker Consistory of the Scottish Rite, BPOE 1829 of Enterprise, Hurricane Creek Grange, Oregon State Grange, and American Legion Post 18 of Enterprise. All of these reflected his active involvement in the community. 

Before his death, he was also affiliated with the Joseph United Methodist Church. Victor Michael Crow's journey came to an end on October 29, 1975, just four and a half years after marrying Ariss. His funeral, held at Joseph Methodist Church, witnessed officiants R. E. Hefty and Rev. Wm B Watson, with Masonic Committal services by Standley Lodge #113 AF & AM. Victor found his resting place in Lostine Cemetery, leaving behind a legacy resonating with those who knew him.

His story lives on, documented through various sources, including newspapers, certificates, and media archives. Victor's life, marked by triumphs and tribulations, remains a testament to resilience, love, and the enduring spirit of those who face life's challenges head-on.

Image by Moshe Harosh from Pixabay