Two Pulitzer Prize winners reveal their use of AI

On May 6, 15 pieces of journalism were awarded the prestigious Pulitzer Prize for journalism. Among them, “Missing in Chicago” was the series that won this year’s Pulitzer Prize for local reporting. The series exposed the failures of the Chicago Police Department (CPD) in investigating missing or murdered black women. Published by the nonprofit journalism organizations City Bureau and Invisible Institute, the series was years in the making with the help of a machine learning tool called Judy.

The prestigious Pulitzer Prize for journalism. Photo: NiemanLab

“We used machine learning to analyze the text in police files, especially the types of documents that contain inside stories,” said Trina Reynolds-Tyler, director of data at the Invisible Institute, who shared the Pulitzer Prize with City Bureau reporter Sarah Conway.

Reynolds-Tyler began developing Judy in 2021 as part of a project at the Invisible Institute to process thousands of CPD misconduct records. The records spanned from 2011 to 2015 and were used to create training data for Judy.

As a result, Judy brought 54 allegations of police misconduct related to missing persons during that four-year period, as well as the pain of families whose loved ones have gone missing in recent years. Judy showed that these cases were part of a history of systemic failures by the CPD.

In the international journalism category, the New York Times (NYT) won a Pulitzer Prize in December 2023 for its coverage of the war in Gaza. The team trained a tool to identify craters left by 907 kg bombs – one of the largest bombs in Israel’s arsenal.

The NYT used this tool to review satellite imagery and confirmed that hundreds of those bombs were dropped by the Israeli military on southern Gaza, particularly in areas marked safe for civilians.

“There are a lot of AI tools that are basically just tools with strong pattern recognition capabilities,” says Ishaan Jhaveri, a reporter with the Computational Reporting Methods team. He explains that if you need to look through a huge amount of data for an investigative project, you can ask an AI algorithm to help. In this case, AI helped find and identify bomb craters from aerial photos.

As a result, the investigation team found that as of November 17, 2023, there were more than 200 907 kg bomb craters in southern Gaza. The NYT noted: "It is possible that more of these bombs were actually used than what is recorded in our report."

“We use AI effectively because this is the kind of task that would be too time-consuming if done manually, affecting other investigative work,” said Jhaveri.

Ngoc Anh (according to NiemanLab)

Source: https://www.congluan.vn/hai-chu-nhan-doat-giai-pulitzer-tiet-lo-viec-su-dung-ai-trong-bai-bao-cua-minh-post294913.html