I should add, “…and With Absolutely No Judgment”, because AI is not a sentient being and has neither intelligence, ethics, nor common sense.
I’ve put off writing about AI, but several recent experiences have convinced me that too many people are trusting AI without understanding either its strengths or its dangers. That made me realize that I absolutely MUST write this series.
Here’s the challenge, though, and the quandary, which is why I’ve hesitated.
I can’t talk about the good, without talking about the problems and abject failures. I can’t do the reverse either, because there’s absolutely an upside. Plus, AI is getting “better” every day. Better is subjective, depending on how AI is applied.
I’m neither an AI evangelist nor a doomsayer. I’m a cautious practitioner.
Like it or not, AI is here to stay, and it’s already embedded in your life in ways you might not realize or recognize.
AI has great potential for good, helping us in our daily lives. It also has an equivalent potential for evil. There’s a very dangerous aspect of AI, and you absolutely need to be aware so that you can take steps to protect yourself.
That said, AI provides extremely useful tools…under some circumstances. I use it for something almost every day – but NOT to write my articles. These are my words. Yes, I do sometimes ask AI for input, and I’ll share how I balance my work and my words with AI assistance – like creating the graphic in this article.
AI is really about education and balance.
To achieve that, I’m writing a four-part series that will be:
- Encouraging but not advocating for AI
- Friendly rather than alarmist
- Skeptical and vigilant rather than anti-AI
- Educational rather than preachy
- Focused on critical thinking
- Warning when necessary
I’ve been working with AI since the beginning in a very restricted, measured way. I use AI regularly, tactically, and cautiously, with huge guardrails. I took the original classes from Mark Thompson and Steve Little, AI experts that I absolutely trust, to learn how to use AI both productively and safely. That was a couple of years ago, and a lot has changed since then. I make it a priority to stay current. We’ve been growing as a community ever since, celebrating our successes and analyzing the failures.
Mark and Steve say:
- Know your tool
- Know your limitations
- Know your data
I would add
- Know your subject
- Know what can (and will) go wrong
You absolutely, positively must check and verify everything AI tells you, without fail.
Being trusting and over-confident is a fool’s errand and assuredly will come back to bite you, sooner than later. It’s essential to be hypervigilant.
In a nutshell, AI is a wonderful servant, although sometimes it has an attitude and doesn’t listen to directions well, but it’s a terrible authority. AI, much like my teenagers used to do, fibs very convincingly and with impunity.
As the adults in the room, it’s up to us to always monitor and check AI output – and learn to recognize it when others use it as well.
That’s the purpose of this series. I’ll be combining my computer science background and genetic genealogy expertise with a couple of years of hard knocks in the AI arena to help everyone be safe and effective. I’ll be sharing successes and failures, good examples, and do I EVER have a great bad example for you.
Articles will include:
- All About AI – What is AI and How Does It Work?
- AI Assistants – The Good, the Bad, the Ugly and the Clandestine
- AI and Genealogy – Brick Walls, Breakthroughs and Blunders
- The AI Threat Landscape – Evil, Dangers and You
My Dad used to tell me, “You don’t have to roll in every mudpuddle that the rest of us have rolled in just to come out the other side saying it’s wet and it’s muddy.”
Some lessons are better learned by someone else going first.
Technology changes, but human nature doesn’t. The tools may be new and revolutionary, but the risks of overconfidence, misplaced trust, and wishful thinking are as old as humanity itself. So come along and join me for the next article, where I’ll share what finally pushed me over the edge to write this series.
_____________________________________________________________
Share the Love!
You’re always welcome to forward articles or links to friends and share on social media.
Subscribe!
If you haven’t already subscribed, it’s free. You’ll receive an e-mail whenever I publish by clicking the “follow” button at the top of the main blog page, here.
Help Keep This Blog Free
I receive a small commission when you click a vendor link in my articles and purchase that item. This does NOT increase your price but helps me keep the lights on and this informational blog free for everyone. Please click on the affiliate links in the articles or to the vendors below if you are purchasing products or DNA testing.
Thank you so much.
DNA Purchases and Free Uploads
- FamilyTreeDNA – Y-DNA, mitochondrial and autosomal DNA testing
- MyHeritage DNA – Autosomal DNA test
- AncestryDNA – Autosomal DNA test
- AncestryDNA Plus Traits
- 23andMe Ancestry – Autosomal DNA only, no Health
- 23andMe Ancestry Plus Health
Genealogy Products and Services
- MyHeritage Subscription with Free Trial
- Legacy Family Tree Webinars – Genealogy and DNA classes, subscription-based, some free
- Legacy Family Tree Software – Genealogy software for your computer
- OldNews – Old Newspapers with links to save to MyHeritage trees
- MyHeritage Omni comprehensive “everything included” subscription plan
- Newspapers.com – Search newspapers for your ancestors
- NewspaperArchive – Search different newspapers for your ancestors
My Books
- DNA for Native American Genealogy – by Roberta Estes, for those ordering the e-book from anyplace, or paperback within the United States
- DNA for Native American Genealogy – for those ordering the paperback outside the US
- The Complete Guide to FamilyTreeDNA – Y-DNA, Mitochondrial, Autosomal and X-DNA – for those ordering the e-book from anyplace, or paperback within the United States
- The Complete Guide to FamilyTreeDNA – Y-DNA, Mitochondrial, Autosomal and X-DNA for those ordering the paperback from outside the US
Genealogy Books
- Genealogical.com – Lots of wonderful genealogy research books
- American Ancestors – Wonderful selection of genealogy books
Genealogy Research
- Legacy Tree Genealogists – Professional genealogy research
Share this:
Discover more from DNAeXplained - Genetic Genealogy
Subscribe to get the latest posts sent to your email.
Do you see archives such as Kew, NRS, NAI, PRONI and others, using AI to interpret and digitally archive their acient texts? I read somewhere several years ago about the Archives building in Glasgow or Edinburg (maybe both?) where the roof is literally leaking on ancient documents. Appalling if true. It would be nice to get these texts digitized and archived before they are no more!
I wish I knew. I surely hope so. If not directly, even working with someone like FamilySearch to preserve what they have digitially. The digitally archiving part is separate from the AI part – and that should be a priority. You can AI later if you have the images – without them, you’re sunk.
I’d love to go to any one of them and volunteer my services to at least get the scanning done. It just makes me nervous to think that these precious documents can be snuffed out with a snap of the fingers (PROI).
Contact them.
Yeah, that’s been on my bucket list for about ten years. Time to get off the you know what!
I can’t wait to read your series!
Thank you for pointing this out! A.I. research on the family tree can lead to tons of false positives and negatives, and the certainty of the responses really can throw you off track if you aren’t careful.
I used the Google AI a week or so ago to try to locate a reference to a land grant, seeing if it could give me an easy citation without going through indexes to locate it. It gave me a citation, but that volume and page number were for a completely different person. When I called the AI on it, it’s response was “I must have been hallucinating”.
As an experiment, I tried other searches just to see how bad (or maybe good) the AI was. When I asked for any records it could locate of a person in a specific county in a specific time range (there was only a father and son of that name in the county) it gave me a wife and how he had served in Lord Dunmore’s War. When I told AI that the wife it had named was for a different person of the same name, in a different state, it’s response was sorry, I mixed up two different families. When I asked for a specific reference showing this person was in Lord Dunmore’s War I checked the responses it gave me. While those references do list two brothers of the person who were involved, there is no mention of the person I asked about.
AI is only as good as the programmer and the search criteria. GIGO!
And what it’s trained on.
If it’s relying on Ancestry trees, we’re in big trouble!!
Hi Roberta,
As with any other tool, verify, verify, and then verify your verification.
Hi Roberta,
I’d like to tell you about my experience with AI. My dad was the 3rd of fojur sons of my grandparents. The first son, was premature and only lived 28 days. The second son was stillborn. My dad was the 3rd son. The 4th son was stillborn. When I tried to enter this information on my tree in My Heritage, AI accepted the dates and information for the first child and my dad. For the second child I put in the birth and death on the same day in 1905 and for the 4th child I did the same in 1909. AI said that was wrong. There were only 3 sons and the second son was torn in 1905 and died in 1909. AK wouldn;t let me correct it. When I called My Heritage Customer Service and explained the situation, I was told by the person on the phone that I was wrong and AI was always right. That’s when I cancelled my subscription to My Heritage and went back to Ancestry.
What you encountered isn’t AI. It’s plain old-fashioned date-check coding like we’ve been doing for decades. So, it’s fine to be unhappy if you wish, but this one isn’t due to AI. People are equally as unhappy with Ancestry for other issues that also aren’t AI. It’s difficult to know what is and is not AI today, but computers have been “helping us” for decades now. Sometimes accurately, sometimes not. However, AI, as we know it, has only been on the scene now for 2-3 years. The date field is programmed to accept only certain things, within certain parameters. That’s how number fields prevent text from being entered, or from entering years in the future, for example. However, if it did something that appeared to be “thinking,” like making a life-graphic about your ancestor that included the births and deaths of those children – that’s generative AI. If you asked google which disease was the most prevalent epidemic in the 1600s in New England, and it gathered up information, then processed it into a chart to provide you with an “answer,” that’s a form of AI. If it just lists resources for you, that’s not AI. I’ll explain more in the next article.
The point about AI having ‘no judgment’ is spot-on. We need to treat it as a tool, not an authority, especially in fields like genetic genealogy where accuracy matters.
Excellent, Roberta, too many people go absolutely gaga when they hear something has come out of some computer program such as AI and believe it must be the revealed truth. I’m still trying to work out what I can use it for. It did one picture for me, for an inside cover. It certainly can’t “analyse my book” as I was promised, it runs out of grunt after about 20 pages.
Absolutely. I understand that Claude is better for long documents. I recently asked ChatGPT to help ms reorganize a document. 49 pages. It said it could. We agreed on the organization. Then, finally, after making lots of excuses, it finally admitted it couldn’t. Wasted probably 2 hours for nothing. I could have had it done myself by that time.
I am looking forward to this series! I do expect however that the excellent way you approach the topic will ALSO be useful to those interacting with non-AI genealogy software such as WATO, BanyanDNA, etc. It’s the same thing even if the specifics are different: Know your tool, Know your limitations, Know your data, Know your subject, Know what can (and will) go wrong. All of these tools are assistants – even incredible assistants – but they are not the genealogist.
You are absolutely right. That sort of goes to the question if “what is AI anyway?”
Roberta, your timing is spot-on. I recently learned about a free, on-line AI Basics course that sounded worthwhile, but after reading about your new series, I’ll hold off on that course. Yours is the voice I know I can trust!
That course is probably a good idea too. Finding a course is one if my recommendations.
Also, thank you.
Roberta, I am looking forward to this course. I surely don’t understand how AI can help me. Scares me somewhat! Janet Seegmiller
Very much looking forward to this series. I very cautiously use AI for specific tasks I feel comfortable verifying… I shudder when I see it being advocated as a perfect research assistant, especially for people new to genealogy and possibly new to research.
You are taking EXACTLY the right approach.
I’m looking forward to this set of blog posts. I’ve used ChatGPT Plus to transcribe handwritten deeds and other documents. Have to check everything, but ChatGPT Plus was much worse in transcribing my ancestor’s request for a pension for his Revolutionary War service than it has been in transcribing deeds (perhaps because of the boilerplate language?)
I like your style, your content. I know nothing I should know about AI. From the onset of AI, it has seemed to me to be like the day the first hairy humanoid picked up a big rock and wondered what he could do with it. Surely he could discourage monkeys when they go too close and he shouldn’t smack himself in the face with it, but he could make his brother in law think twice. Dangerous but helpful. And such has been the history of hominids like us and tools. Your piece reminds me I need to learn how not to smack myself in the face with it.
What a great analogy. And thanks for the chuckle!