Ancestry used to need 9 months to scan census records and make connections for users. AI cut it to under 9 days.

Ancestry, a leader in genealogy services, is currently beta testing a groundbreaking AI feature that promises to transform how users connect with their family histories. The company boasts an impressive collection of over 65 billion records from more than 80 countries, amounting to about 10,000 terabytes of data, according to Sriram Thiagarajan, Ancestry's Chief Technology Officer. Founded in 1983, Ancestry has long been dedicated to helping individuals trace their roots through an extensive array of records, including birth, death, marriage, and military documents, as well as immigration files and newspapers. Collaborating with institutions like the National Archives, Ancestry has amassed a wealth of information. However, managing such a vast database has posed significant challenges. Thiagarajan recounted that when he joined Ancestry in 2017, the organization was just beginning to explore the capabilities of AI and machine learning. "We were seeking a more efficient method to digitize the content we acquire globally," he explained. Previously, the process involved scanning records and then outsourcing the labor-intensive task of manually indexing the data, which was time-consuming and costly. For instance, digitizing the 1940 census took nearly nine months and was ten times more expensive than today's methods. In a bid to innovate, the Ancestry team began applying AI and computer vision techniques to automate the digitization process. By 2021, they had developed proprietary handwriting recognition technology that significantly reduced the time required for processing records to under nine days. Thiagarajan emphasized the importance of maintaining accuracy in AI-generated results, stating that human oversight remains a vital part of the process. Ancestry's commitment to accuracy is evident in their development of automated systems designed to streamline checks on AI outputs. Thiagarajan noted, "We aim to ensure the results produced by AI are grounded in truth and facts." He also expressed that the advancements made in recent years have markedly improved their operations. In addition to backend improvements, Ancestry is rolling out user-friendly features, including a handwriting recognition tool and an AI assistant set to launch in 2024. The company is also testing an exciting new feature called Audio Stories, which will allow users to transform their records into narrative audio formats. "Our AI understands the context of printed material, images, and handwritten notes, linking everything into a cohesive story," Thiagarajan explained. While there is no official release date for Audio Stories, Ancestry is already looking ahead, envisioning a future where storytelling can incorporate sight, sound, and motion.