AI can never tell you your great-great-grandfather's story


AI has been able to perform a lot of tasks really well these days. However, one thing that it will **never** be able to do is tell you who your ancestors are. You might think that Ancestry.com, FamilySearch, or FindAGrave could have some AI tool attached to them, allowing us to ask them questions about our family tree. This is entirely possible and would be quite useful, but one glaring issue throws a wrench into this entire idea: they can be wrong.

Yes, Ancestry.com, FamilySearch, and FindAGrave are capable of being wrong, or incomplete to a misleading degree. This is because they allow unsubstantiated claims about people that users can add without question. Nothing is validated. As a result, when researching your own family, you **cannot** take anything you find on these sites as fact. Now, they _are_ excellent tools at corroborating evidence, and I'm grateful that they help people collaborate, but they are not evidence in themselves. I use them as tools to piece together a greater narrative that they can never provide on their own.

One example of this happening to me was when I was researching my great-great-grandfather [Robert Edwin Crawford](https://www.crawfordlineage.com/person/0e2f186b-55b7-496f-8541-2a10b40f2863). He's a very mysterious man in my family tree, because he was divorced by [Bertha Zimmerman](https://www.crawfordlineage.com/person/8a783f16-9ff4-448e-8234-9221aa196e89), and disliked by almost everyone in the family after that. Most people in my family have little knowledge of him. Today, I'm highly interested in who he was. I started by gathering as many vital statistics (which constitute as verifiable evidence) as I could:

1. I received a copy of his death certificate from the State of Nebraska. This contained:

   - His parents names, [James Huston Crawford](https://www.crawfordlineage.com/person/feb49296-d16e-4487-b44d-c5b4f09e5a5d) and [Wilnetta Julia Coon](https://www.crawfordlineage.com/person/6d336512-b995-4c84-b202-64741760940d), establishing identity
   - The name of a deceased spouse, "Minnie Crawford"
   - The reporter of Robert's death was "Mrs. Ernest Crawford," his brother's wife

With this data, I had a problem. There was a mixed, incomplete amount of information out there. FamilySearch said Robert had a wife named "Mary Bowlby," Ancestry.com thought he had a wife named "Mary Spears," and FindAGrave thought that Minnie had multiple husbands, but her obituary did not corroborate this. Very confusing. What really happened? Could we toss an AI tool into the mix and have it figure this out for us? I don't think so. Here's why:

## Different Names

Robert Edwin Crawford's middle name is spelled differently depending on where you look: Edwin or Edward, typically. His name is also spelled simply "R. E. Crawford," Robert Crawford, or even just Edward/Edwin Crawford. How can AI know that these are the same people? I'll admit, a confidence score could help an AI determine this, based on locality, spouse data, and similar pronunciation. But there's a large margin for error here given that there were a million other people with all those combination of names alive at that time.

## Incorrect Dates

Individual contributors on FamilySearch, Ancestry.com, and FindAGrave can't agree on Robert's birthday, date of death, and various marriage dates. Neither can they agree on how many spouses or kids he had. This presents a problem for AI, which can only see this mixture of true and false data, and take it at face value.

I have the luxury that AI does not: Robert's death certificate, and greater heuristics. Looking at all of the data available, I believe Robert Edwin remarried two more times after divorcing Bertha, with each of his other wives passing away.

AI can get better at heuristics and connect the dots on his life events, but it will never be _certain_, which is something I require. If all death certificates were electronically available to AI, would that solve the problem? No, because prior to the 1900s, most states didn't even _record_ death certificates. That data literally does not exist in that form. Could AI figure it out via newspaper obituaries? Not everyone paid for an obituary. As an example, Robert Edwin's [grandmother](https://www.crawfordlineage.com/person/00c70b8a-4da4-4af6-bfc5-b27a32e0d41d) had an obituary, but not his grandfather, for whatever reason. So how do we know when they died? That leads us to the last reason:

## No Stories

The previous two reasons could genuinely be solved for AI if all data was electronically available, and high inference was used to establish confidence of identity via obituaries, documents, marriages, etc. in newspapers. But there's one thing that AI will never be able to do: know your stories.

There's simply no record of the feelings people had for Robert documented. If we have stories, it helps us toss out extraneous data that expand the locality too wide. Stories help us narrow evidence down, and find alternative sources around a clue that wouldn't otherwise have been attached to a person. AI doesn't know our ancestor's oral stories that have been passed down.

1. AI doesn't know that Robert Edwin, after his divorce from Bertha, tried sending her money and hated that they viewed him as a deadbeat.
2. AI doesn't know that Bertha's father was Robert Edwin's old employer, and despised him and tried to convince people that he was mentally handicapped.
3. AI doesn't know that Bertha really didn't want to divorce him, but followed her family's pressure to find a man with more stability.
4. AI doesn't know that, contrary to Bertha's family's account, he was a blacksmith his entire life and never really left the area. He was very consistent with his work.
5. AI doesn't know that Bertha's father stole all of the wages from [Lorin Lonzo Crawford](https://www.crawfordlineage.com/person/5999039a-efb7-4cb7-b1f9-6d04804b9aba), one of Robert and Bertha's sons, when he was a child working for the coal mines. Which corroborates Robert's confusion as to why Bertha never got support money from him.
6. AI doesn't know that _someone_ out there loved Robert, even if my own side of the family did not. I know this because Robert lived with his son Lorin for a time.
7. AI will not be able to recognize Robert Edwin. I've never seen a photo of him, but I know that there is one out there, and I'll know it's him because I know what all of his [children](https://www.crawfordlineage.com/person/7e4764cd-f9f1-4a5b-ae73-a381445752f0) looked like, and their [children](https://www.crawfordlineage.com/person/9e9f1aee-0956-49a1-8fe6-aa820bb79d4d), and their [children](https://www.crawfordlineage.com/person/d1c589c3-d35c-48f0-af1f-0040da00a53f), and [me](https://www.crawfordlineage.com/person/b14745c4-3d98-4bc3-a4d0-7e973c8a88ac).

## Conclusion

Talk to your relatives about your own family, listen to and remember the stories they tell. Once a generation passes away, everything they knew will suddenly disappear, and AI will **never** be able to figure it out. If you don't care about genealogy, do it for the generation that will care. Some day, someone is going to ask about their family, and the time you put into it now will be a blessing to them later.


The Value of Obituaries


I've been pursuing membership with the [Sons of The Revolution](https://www.sar.org/) organization, because of my direct ancestor [Benjamin Crawford](https://www.crawfordlineage.com/person/672418b7-1163-4ad7-bcbc-83269fb9f4f4)'s military service in the revolutionary war. This requires vital records proving every generation up to the ancestor in question. Vital records are birth, marriage, or death records that help establish the parents and children of an individual. This is all great, however when you start delving into the 1800s and earlier, you run into a problem. Nobody reliably has these types of records on file, and the government wasn't typically keeping track of them anyway. So, how do you prove your lineage? You look at obituaries.

You might be wondering if census data can also be used as a vital record, because it's really easy to find. But the reality is that, especially in the 1700s, they don't say who's in a household, only the head of the house. This doesn't establish any type of relationship. Obituaries, on the other hand, can be a goldmine of information. They count as vital records because they often list surviving family members, and the parents of the deceased. So how do we find them?

You have a couple options:

1. Hope that your family kept old newspaper snippets
2. Pay money to the [https://www.newspapers.com/](newpapers.com) website
3. Figure out where someone died, and search the local archives

I prefer the last option, because it's free. But what if you can't find any obituaries for a specific person? You need to start looking for adjacent obituaries. For example, I've had a difficult time finding any evidence that [James Huston Crawford](https://www.crawfordlineage.com/person/feb49296-d16e-4487-b44d-c5b4f09e5a5d)'s father was [Robert Crawford](https://www.crawfordlineage.com/person/46b9914d-3b7a-467a-886f-1ddd35f0279b), apart from census data and old family stories. James doesn't have any obituaries available as far as I've been able to find. But he had a lot of siblings, and one in particular was very well documented. His name was [William Chambers Crawford](https://www.crawfordlineage.com/person/45ed1f17-5409-4b0e-914f-a2da85d39f78), and I found a very detailed obituary in the Tama County newspaper archives:

![](https://drive.google.com/uc?export=view&id=14-hq2RuXXPIIMdVAxqyKBaG74ayz61Kc)

[Check out the whole article here](https://tamacounty.advantage-preservation.com/viewer/?k=william%20chambers%20crawford&i=f&by=1922&bdd=1920&d=01011922-02281922&m=between&ord=k1&fn=toledo_chronicle_usa_iowa_toledo_19220216_english_1&df=1&dt=3&cid=2711)

The key to this obituary is that it specifically mentions that his mother and father was Robert and Elizabeth Crawford, and that he had a surviving brother "J H Crawford" who was living in Washington State at the time. This gives us enough evidence to prove that James Huston Crawford was the son of Robert Crawford. This is a great example of how obituaries can be used to prove lineage.

![](https://drive.google.com/uc?export=view&id=1hQfVWF9LThXr1S5BZVUvXd2eDoToqEki)

So, if you're feeling stumped trying to prove out a relationship, check out the obituaries of the siblings, parents, and children of the person in question, and you might find the exact evidence you're looking for.


The Crawford Genealogy Blog

AI can never tell you your great-great-grandfather's story

The Value of Obituaries