Divining data: How AI invigorates The New York Times’ approach to investigative reporting
- The New York Times is integrating artificial intelligence into it's newsroom operations, enhancing investigative reporting through advanced AI applications.
- Seward, formerly CEO and editor in chief of Quartz, spoke at WAN-IFRA’s recent Congress in Krakow about how the newspaper's AI applications are becoming essential for journalists.
- The AI Toolkit for Investigations addresses common challenges journalists face when dealing with vast amounts of data.
NYT’s AI Toolkit Enhances Investigative Reporting
Updated May 31, 2025
The New York Times is integrating artificial intelligence into it’s newsroom operations, enhancing investigative reporting through advanced AI applications. Zach Seward, the NYT’s editorial director of artificial intelligence initiatives, leads the AI Issues team, a multidisciplinary group focused on applying AI tools across various newsroom functions.
Seward, formerly CEO and editor in chief of Quartz, spoke at WAN-IFRA’s recent Congress in Krakow about how the newspaper’s AI applications are becoming essential for journalists. The team’s work spans internal workflows, prototyping, and exploring future AI uses.
The AI Toolkit for Investigations addresses common challenges journalists face when dealing with vast amounts of data. This toolkit includes bias-based search, data mining, dataset augmentation, and end-to-end verification.
according to Seward, large language models (LLMs) are invaluable for analyzing extensive document and video collections, providing journalists with a “superpower” to search data in previously unfeasible ways. The AI toolkit helps journalists overcome the challenge of unpacking massive reams of data.
Key Components of the AI Toolkit
Vibes-based search,also known as semantic or vector search,uses vector embeddings to identify semantically similar content beyond exact keyword matches. This helps journalists uncover connections and patterns that conventional search methods might miss. “By using semantic search, we were able to find a much, much wider swath of examples,” Seward said.
Diving for pearls is a data extraction tool that uses AI, guided by journalist expertise, to extract insights from large volumes of content.The NYT used this to analyze over 500 hours of leaked video from an election interference group. The AI transcribes videos into text, which reporters then analyze.
Augmenting datasets involves using optical character recognition (OCR) to analyze complex document sets,including handwritten notes. Seward noted that recent advancements in foundational models have considerably improved OCR capabilities, enabling complex analysis of messy datasets. The NYT has also developed tools for monitoring online content and screening individuals for investigations.
End-to-end verification is crucial, as Seward emphasized that the NYT never trusts an LLM without verifying its output. The verification tool links AI-generated insights back to primary sources, requiring journalists to review original material before publication. ”It’s crucial that we design these systems in a way that the original material is accessible and tied directly to the analysis,” Seward said.
“We now, as journalists, have a capability to search through data sets in ways that previously were not possible and really give our journalists a whole new superpower.”
Zach Seward, The New York Times
The ‘Cheat Sheet’ Tool
The NYT is developing a “cheat sheet” tool to help journalists make sense of large