To say that Ancestry's database is huge is an understatement.
“We have collected over 65 billion records in over 80 countries,” Sriram Thiagarajan, the company's Chief Technology Officer and Executive Vice President of Products and Technology, told Business Insider. “To give scale, it's about 10,000 terabytes of data on the platform that we use to provide discovery to our users.”
Founded in 1983, the Utah-based genealogy company collects records to help people unearth the roots of their families.
These records include birth, death, marriage, census, military, land, immigration and newspapers. Ancestry, which also provides consumer DNA test kits, works with agencies such as the National Archives and Records Management to collect that data.
However, trobs of that size have difficult caveats. Let's sort it out.
Chiagarajan said Ancestry is learning to leverage AI and machine learning to make Herculean tasks easier.
Streamline with computer vision
When he joined the Ancestry team in 2017, Chiagarajan said the company was just beginning to explore AI and machine learning.
“We were trying to figure out an effective and efficient way to digitize content we retrieve from around the world,” Chiagarajan said.
The ancestor scanned records, outsourced operations and manually indexed them to manually index important fields. That information was uploaded to Ancestry's database before the software program established relationships between people, locations, or other categories.
“Around 15 or 20 years ago, when we digitalized the 1940 census, it took about nine months to do it manually at 10 times the cost,” Chiagarajan said.
This led to the ancestor team searching for answers.
Related Stories
Business Insider tells the innovative stories you want to know
Business Insider tells the innovative stories you want to know
“We said, 'Why not apply computer vision AI technology to automatically digitize content without manual intervention?” Chiagarajan said, “Fast-forward to the 2021 time frame, using its own handwritten-recognized computer vision technology, bringing it to the market within nine months to nine days at some cost.”
Chiagarajan said that his ancestors will expand on the technology to process other record types, but humans will still review AI results “if necessary.”
“We've built automated controls and systems that certainly reduce the time we need to spend checks,” he said. “We want to be particularly careful about making sure that what we produce using AI is based on truth. It is based on fact.”
Chiagarajan added, “The degree to which we do now has definitely improved compared to a few years ago.”
“At the end of the day, when consumers come to our platform looking for stories about their ancestors, we want to tie them together with records of finding them,” he said.
Ancestry is beta testing new AI features
In addition to implementing AI in the backend, Ancestry has rolled out several features aimed at users, including handwriting recognition tools. We began testing AI assistants in 2024.
Recently, Chiagarajan told Business Insider that Ancestry has beta-tested on AI-powered features called Audio Story, which allow users to turn records into story audio.
“Our AI can understand the context between printed materials, images and handwritten stories and connect it all to the story,” Chiagarajan said.
There is no official release date for Audio Story yet, but Ancestry is already thinking beyond audio.
“I want to go down the road and be able to add more visual, sound, movement and video storytelling,” Chiagarajan said.
