23 and me could uses AI to create stories about your long lost loved ones

Each day I’m thrilled to share with you the mind-blowing potential of AI applications in new industries. Some days you can try out the technology yourself like in today’s story. Thanks to systems like chatGPT, 23 and me is able to create stories about the people who lived before you with astonishing accuracy. And while this consumer application might not be that popular in the narcissistic bubble of the USA, it definitely is a use case that will change how you think about your history. You can try the DIY version yourself and see how it makes you feel. I was a enriching experience for me.

I don’t know much about my Great GranDad (GGD). I’ve asked my father about him a few times, and he can tell me a couple of stories about what he remembers and what his father told him. I love the oral tradition of passing down stories of loved ones to kids, and so learning about the history of my lineage has been intriguing to me. ChatGPT doesn’t know anything about my family or who I am; however, it has a very good textual understanding of history and cultural contexts by region. I wondered if ChatGPT could generate stories about what my GGD might have experienced when he was in his early 20s. Immediately I began to wonder how accurate GPT could be if it was generating stories about my GGD with some very simple information that I knew about him. So rather than starting with him, I began with myself.

In 1997, I worked for UNSCOM in Baghdad translating Arabic in the search for weapons of mass destruction. I knew what my days were like there, so I posed this prompt to see how accurate it could be for me.

Tell me a story about what an Arabic translator’s day might have been like when they were working as an inspector for UNSCOM in Baghdad in the late 1990s. — The key to this prompt is to pinpoint a specific moment in your history accurately; otherwise, your responses may be off.

“It was another intense day for Farah, an Arabic translator and inspector working for the United Nations Special Commission (UNSCOM) in Baghdad during the late 1990s. She arrived at the UNSCOM headquarters early in the morning and was immediately briefed on the day’s inspections. Her first task of the day was to inspect a suspected missile site on the outskirts of Baghdad. Farah joined a team of UNSCOM inspectors, who were tasked with collecting samples and examining the site for any signs of prohibited missile activity. Farah’s role was critical in ensuring that any documentation or communication in Arabic was accurately translated for the rest of the team.
As the day wore on, Farah and her team faced numerous challenges, including hostile Iraqi officials who attempted to block their inspection efforts. Farah was called upon to provide interpretation in tense meetings and negotiations with Iraqi officials, ensuring that the team’s objectives were communicated clearly and accurately.
Later in the day, Farah had a meeting with a group of Iraqi scientists who were involved in the country’s weapons program. The meeting was conducted in Arabic, and Farah was responsible for providing simultaneous interpretation of the proceedings into English. The scientists were reluctant to provide information and were often evasive in their answers, making Farah’s job even more challenging.
As the day drew to a close, Farah returned to the UNSCOM headquarters to prepare for the next day’s inspections. She knew that the work was often challenging and dangerous, but she felt a sense of satisfaction in knowing that she was playing a crucial role in preventing the proliferation of weapons of mass destruction in the region.
Despite the difficulties of her work, Farah knew that she was making a difference and was committed to continuing her work as an inspector and translator for the UNSCOM.”

The response was pretty spot on. ChatGPT was able to approximate what I would have experienced because it took into account a variety of variables at the time, such as the tensions between Iraq and the UN, our prior war experience, the culture in Iraq, and so on. I was quite happy with how close it was, and I was now eager to learn about my GGD.

Next, I tried to estimate the year when my GGD was 25, because I don’t remember when he was born. If I ask my mom, she might talk for hours. So, I made some more approximations. Now, let’s ask ChatGPT to break down the problem down for us:

“Give me the order of operations that you would have to do to provide me with the approximate year my GGD was born. take into account the average age that men had kids was about 23 and my dad is 83.”

Sure, here are the steps you can follow to estimate the approximate year your GGD was born:
1. Subtract your father’s age from your age to find the age difference between you and your father. For example, if you are currently 47 and your father is 83, the age difference is 36.
2. Divide the age difference by 2 to find the approximate age of your father when you were born. For example, if the age difference is 36, your father would have been around 18 years old when you were born.
3. Add your father’s age at your birth to your own age to find the approximate age of your grandfather (your father’s father) when your father was born. For example, if you are currently 47 and your father was around 18 when you were born, your grandfather would have been around 65 years old when your father was born.
4. Subtract 23 from the age of your grandfather when your father was born to estimate the year your GGD was born. For example, if your grandfather was around 65 when your father was born, you would subtract 23 years from that to estimate that your GGD was born around 1897.
5. Adjust the estimated year based on any known information you have about your family’s history or your great-grandfather’s life events.

That doesn’t look like how I would do it, but let’s just give it a shot.
Now that I have the list of operations, I can ask ChatGPT to follow those instructions and tell me what year it was when my GGD was 25.

Can you do that and tell me what year it was when my GGD was 25?

1922. Great. Now we have a starting point. I only know a few things about GGD:
1. He worked on threshing crews going from north to south in the Midwest.
2. He had a pretty cool pool cue made of four pieces, which was nice.
3. He spent some time in Chicago before coming back to North Dakota and homesteading on the Missouri and Yellowstone.
That’s enough to get into some of the details of his life. Now, let’s engineer a few prompts to get the information that we need.

Write me a list of procedures you will need to do to tell me a story about the day in the life of a man who in 1922 worked on the traveling threshing crews, had a very nice pool queue that came apart into pieces, spent some time in Chicago and was boisterous and german. Add to the process: tell about the context of the news at the time.

This prompt is getting pretty large. I find that working with GPT that you need to attempt to stay below 2 instructions in each interaction. Here’s was the response:

sure, here are the procedures you can follow to tell a story about the day in the life of a man who worked on the traveling threshing crews in 1922, had a nice pool queue, spent time in Chicago, and was boisterous and German, while also providing some context of the news at the time:
1. Research the historical context of the United States in 1922, including major events, social issues, and cultural trends.
2. Gather information about traveling threshing crews, including their work, lifestyle, and typical crew members.
3. Create a character sketch of the man, including his name, age, background, personality traits, and goals.
4. Write a narrative that describes a typical day in the life of the man, including his work on the threshing crew, his interactions with other crew members, his love for his pool cue, and his time spent in Chicago.
5. Incorporate details about the historical context of the time into the narrative, including references to major events or social issues that would have been relevant to the character and his experiences.
6. Edit and revise the narrative to ensure that it flows smoothly and accurately reflects the character and the historical context.
7. Finally, read the story aloud to ensure that it is engaging and captures the attention of the listener.

OK, this is pretty good. As a side note, it really bothers me when ChatGPT is too accurate. It has done two things so far in this chat that creep me out. It guessed my age and knows that it can talk to me. — I have the application called Talk to ChatGPT, so it does actually read aloud to me. Number 7 above is accurate.

OK, execute that procedure. My GGD’s name was Adam.

Sure, here’s a story about a man named Adam who worked on traveling threshing crews in 1922, had a nice pool cue that came apart into pieces, spent time in Chicago, and was boisterous and German:
It was another hot summer day in 1922, and Adam was up early, getting ready for another long day on the threshing crew. He lived in a small town in the Midwest and traveled with the crew from farm to farm, threshing wheat and other grains for the local farmers.
Adam was a tall, burly man with a thick German accent and a boisterous personality. He had been working on the threshing crew for several years, and he loved the camaraderie of the other crew members and the satisfaction of a hard day’s work.
As he packed his bag for the day, Adam carefully placed his favorite possession — a beautiful pool cue that came apart into pieces — into its leather case. He had acquired it during a visit to Chicago a few years earlier and treasured it as a reminder of his time in the city.
The crew loaded up their equipment and headed out to the first farm of the day. Adam worked hard, sweating in the sun as he operated the threshing machine and helped to load the grain onto the wagons.
During a break, Adam pulled out his pool cue and challenged some of the other crew members to a game. He played with skill and enthusiasm, his boisterous laugh ringing out across the farmyard as he sank shot after shot.
As the day wore on, Adam grew tired and his muscles ached from the physical labor. But he pushed through the fatigue, determined to finish the job and move on to the next farm.
In the evening, the crew stopped at a local tavern for a drink and a meal. Adam regaled the other crew members with stories of his time in Chicago, describing the bustling city and the sights and sounds that had captivated him during his visit.
As the night wore on, the crew grew tired and headed back to their temporary campsite, ready for a good night’s sleep before the next day’s work. Adam settled into his bed, his mind filled with memories of his travels and the thrill of the game of pool.
Meanwhile, in the wider world, 1922 was a year of significant events and social change. The country was still recovering from the aftermath of World War I, and tensions were high between labor unions and business interests. Women’s suffrage was a major issue, and the Roaring Twenties were just beginning to bring about significant cultural and social changes.
Despite the challenges of the time, Adam and his fellow crew members worked hard and took pride in their work, carrying on the traditions of the farming communities they served.

This is just a starting point. You can refine the story by asking ChatGPT to do a variety of character projections and put your grandfather in some generic situations.

That might include:
Day in the life: What sort of difficulties would he have experienced?
Day in the life: What did he do to date?
Day in the life: Tell me about how Adam was able to homestead in the 1920s.

This exercise is a wonderful way to get an idea of what life was like for your ancestors. this could be easy for a company like 23andMe since they have a large database with the keys to historical context like locations and dates.. They could generate many historical stories about relatives as a way to connect families with histories. It would also be nice if users could add some notes to the application and it would use those notes to approximate with more flair.

In summary: It’s also pretty obvious how this could morph into talking to your relatives, using your relatives as your AI companions or loading them into games as part of the adventures. The suspension of disbelief works well because you don’t know the personalities of those from way back when. A generated story or personality could easily suffice. Enterprise organizations could prime ChatGPT using chaining for this task over the new ChatGPT 3.5 turbo since it is very limited.

Thanks for reading. This story is part of a series regarding the near future of business applications and how they will change our society. I’m conceptualizing these applications, which are very much possible, using my understanding of data science, ML and AI methods, available technologies and market needs. The topics in this series will range from general AI integration in industries to highly technical business cases such as intelligent order management systems or predictive spent categorization.

Join me on this journey to explore the limitless potential of AI in business applications, from art to technical solutions. Your part in this discourse will help everyone as we learn to understand where we are going and the dangers of the powerful capabilities of AI.