Compiled by the editorial desk with reference to official statements, public filings, and industry data.

In November, the Knight Foundation funded a new initiative at Quartz, the digital news outlet, to explore how machine learning could aid journalism. Today, that project—Quartz AI Studio—published its first story, offering a glimpse into how algorithms might help reporters analyze dense public documents.

The story focuses on Lyft's initial public offering (IPO) filing, specifically the section listing risk factors. Quartz reporters trained a machine learning algorithm to scan that section and flag the most unusual or distinctive concerns mentioned by the ride-hailing company.

The algorithm's output highlighted a mix of expected and surprising worries. Beyond typical concerns about public perception, Lyft's executives also expressed unease about how healthcare privacy laws might affect users who take rides to medical appointments. Another flagged risk involved potential cyberattacks on Amazon Web Services, the cloud platform that runs Lyft's operations.

Proof of Concept for Data Journalism

While the resulting article is not a groundbreaking investigation, it serves as a proof of concept for how reporters can use new tools to extract meaningful insights from otherwise dry records. John Keefe, Quartz's technical architect for bots and machine learning, described the approach as an extension of data journalism, aiming to get journalists comfortable using computers for pattern matching, sorting, grouping, and anomaly detection, especially with large datasets.

The project underscores a broader trend in newsrooms: leveraging artificial intelligence to handle tasks that are time-consuming for humans, such as sifting through lengthy regulatory filings. By automating the initial scan, journalists can focus on interpreting and contextualizing the findings.

Quartz's experiment comes amid growing interest in AI's role in media, from automated news writing to content recommendation. However, this initiative is distinct in that it targets the analytical side of reporting, helping journalists ask better questions of public records.

As machine learning tools become more accessible, similar approaches could be adopted by other news organizations to analyze everything from government contracts to corporate disclosures. The Lyft story, while modest in scope, offers a tangible example of how AI can augment human reporting rather than replace it.