Revolutionizing Investigative Journalism with AI

The New York Times has embraced artificial intelligence to unlock investigative stories previously deemed unmanageable due to the overwhelming volume and complexity of data involved. By leveraging AI, the editorial team can now analyze vast datasets and hours of multimedia content, enabling deeper and more comprehensive reporting.

Zach Seward, the AI Editorial Director appointed in late 2023, spearheads this transformative initiative. His role emerged amid a surge of AI-focused positions across media organizations, aimed at integrating AI tools and ethical guidelines to enhance journalistic capabilities and maintain competitive advantage.

How Seward’s Team Empowers Reporters Through AI

Seward leads a dedicated group of eight professionals, including engineers, editors, and a product designer, who collaborate closely with journalists to tailor AI solutions for specific reporting challenges. Their primary focus is on research and investigative applications, which Seward identifies as the most impactful and resource-efficient use of AI in media today.

The team’s approach involves partnering with reporters on individual projects, developing AI-assisted workflows, and then refining these into scalable tools that can be adopted newsroom-wide. This iterative process ensures that AI integration is practical, user-friendly, and aligned with journalistic standards.

Case Study: Uncovering Election Interference Through AI

One notable example involved a reporter tasked with reviewing an enormous trove of 500 hours of leaked Zoom recordings linked to an election interference group. Manually sifting through this material was impossible within the tight pre-election timeframe. The AI team deployed advanced transcription technologies to convert over five million words into searchable text and utilized semantic search algorithms to identify relevant themes and conversations.

Unlike simple keyword searches, semantic search enables the detection of related concepts and nuanced topics, which proved crucial since the group never explicitly admitted to spreading misinformation. This AI-driven analysis culminated in a major exposé published just before the presidential election, demonstrating the power of AI to reveal hidden narratives within complex data.

Developing Custom AI Tools: The ‘Cheat Sheet’ Spreadsheet

Building on these successes, Seward’s team created an internal AI-powered spreadsheet tool named “Cheat Sheet.” This platform allows reporters to select from various large language models (LLMs) to assist with data analysis and research tasks. Currently, dozens of journalists across the Times utilize Cheat Sheet to streamline their investigative workflows.

While Seward declined to disclose specific AI vendors, he confirmed that the newsroom experiments with a broad spectrum of commercial and open-source AI models, ensuring flexibility and adaptability in their toolkit.

Scaling AI for Large-Scale Data Analysis

Another innovative application involves automating the analysis of extensive datasets, such as processing information on thousands of individuals. Seward explained that while humans cannot manually “Google” tens of thousands of names, AI can efficiently perform this task and flag promising leads based on predefined criteria. Although the AI-generated results require human verification, this method significantly accelerates the investigative process.

The team’s philosophy centers on tackling “knotty” and disorganized data under tight deadlines, always with an eye toward creating reusable AI tools that empower the entire newsroom.

Fostering AI Literacy and Collaboration Across the Newsroom

To maximize AI’s benefits, Seward’s team prioritizes ongoing dialogue and education. They have engaged with approximately 85% of the newsroom staff through training sessions, workshops, and an open Slack channel where journalists can ask questions, share insights, and exchange best practices related to AI usage.

This collaborative environment helps demystify AI, addressing individual needs and preferences. Seward emphasizes that AI is a highly personalized tool, with users often encountering “writer’s block” when interacting with chatbots. The team’s role includes guiding reporters to overcome these challenges and harness AI effectively.

Addressing Skepticism and Ethical Considerations

Despite the enthusiasm, Seward is clear that AI is not employed to author articles at The New York Times. Instead, it assists with drafting elements like SEO metadata and headlines. Editorial staff are cautioned to treat AI outputs with the same critical scrutiny as any new source, maintaining journalistic integrity.

The team acknowledges and respects skepticism within the newsroom, focusing on demonstrating AI’s practical advantages rather than promoting it unconditionally. Seward notes that much of their effort involves educating colleagues about the legal and editorial risks associated with AI, fostering a cautious yet optimistic mindset.

He candidly shares his own concerns about potential errors linked to AI, affirming that responsibility for any mistakes ultimately rests with the human journalist, not the technology.