Holocaust researchers use AI
In Israel, researchers are leveraging artificial intelligence (AI) to sift through extensive archives in an attempt to identify hundreds of thousands of Jewish individuals who perished in the Holocaust but are not memorialized in official records.
During the Second World War, the Nazis perpetrated the genocide of over six million Jews, a tragic event commemorated globally on Yom HaShoah, or Holocaust Remembrance Day, observed on Monday.
In preparation for these commemorations, personnel at the Yad Vashem World Holocaust Remembrance Centre in Jerusalem have intensified efforts to uncover information about both identified and unidentified victims by employing their AI-driven software.
Through the dedication of volunteers, details of 4.9 million individuals have been compiled over the years by meticulously examining statements, documents, film footage, cemeteries, and other archival records.
Esther Fuxbrumer, head of software development at the centre, explained to Reuters the difficulty of manually reviewing vast amounts of data without overlooking any details.
She highlighted the significant gaps within their current database of 9 million records, emphasizing that many victims were indiscriminately executed and buried in mass graves without anyone left to document their identities.
The monumental challenge also involves the intricate process of linking individuals to specific dates, family members, and other pertinent details, while also identifying duplicates and cross-referencing accounts.
The AI system, crafted over the last two years to analyze records in multiple languages including English, Hebrew, German, and Russian, is presently undergoing rigorous testing.
Esther Fuxbrumer commented on the efficiency of the technology, noting its speed and precision in processing hundreds of testimonies within a matter of hours.
Esther Fuxbrumer shared insights on the impact of the AI system, stating, “From each testimony, we could extract between six or seven names with comprehensive details, which we could then automatically integrate into our database. About 10% of the names we discovered were already in our database, but the remaining 90% were previously unknown to us.”
She recounted a poignant example where the system uncovered information about Yehudit and Ruth Rosenbaum, twin sisters aged four and a half from Romania, both transported to Auschwitz. Yehudit survived, while Ruth tragically perished.
Fuxbrumer further emphasized the system’s capability to provide additional details about individuals like Ruth, sourced from individuals outside her family, such as someone who encountered her during their time in the camp.
During the ongoing trials, the staff is examining 400 out of their 30,000 testimonies, including numerous extended recorded videos of survivors, some lasting up to three hours.
Fuxbrumer stated that they have already incorporated 1,500 new names, with many more expected in the coming weeks as the system is implemented across all 30,000 testimonies. The next phase of the trial will involve examining diaries.
“We believe that this approach will enable us to uncover numerous stories about victims who were previously unknown, including young children, shedding light on their experiences,” Fuxbrumer remarked.
