Reporting On (and With) Artificial Intelligence

The perception that artificial intelligence has transformed the way we live, work, and interact has been surrounded by hype, misunderstanding, and misinformation. Aiming to address this shortfall, the Pulitzer Center has developed a baseline study to support initiatives that contribute to more impactful AI storytelling.

It all started with a question that kept coming up in team meetings at Salud con Lupa: Why is it so common for older adults to be excluded from Pensión 65 without any explanation? To understand this, they had to look closely at the system that decides who receives support from the Peruvian government. It’s a system that merges databases, surveys households, runs scores through an algorithm… and if the numbers don’t fit, people get excluded.

Pablo Jiminez Arandia leads a cross-border investigation to show how the tech oligarchy is searching almost desperately across the globe for land and natural resources to feed its server farms. He found communities separated by thousands of miles but united in their struggles with extractivism and water stress.

"As we navigated everything from Mindrift’s recruitment pipeline and its star-studded client base, to the overall purpose of this grand operation, we frequently ran into dead-ends. We constantly faced new questions and had to change our hypothesis over and over again. We learned quite early on that we needed to become comfortable with being wrong."

"The story came together in an unconventional, novel way that I believe points to a model reporters, and especially freelance journalists like myself and small newsrooms, can adopt for in-depth data and AI accountability-driven stories, which otherwise would be impossible to conduct."

Niamh McIntyre was interested in the intersection of two harms: working conditions for data workers and people who were subject to excessive surveillance by companies or law enforcement.

What she found linked a network of gig workers in the Global South with Russia's extensive state surveillance system.

Joanna Kao designed her AI Fellowship investigation into how disabled people experience the AI boom in a way that many AI companies marketing accessibility products have failed to do: by centering the individual, their experiences, and their environments.

Working with 20 researchers, AI Accountability Fellow Karol Ilagan teamed up with data specialist Federico Acosta Rainis to spend more than six months examining ride-hailing app Grab’s algorithm and how it impacts consumers and drivers.

A case study on how India’s Bharatiya Janata Party used WhatsApp to spread its message in a small Indian town. We studied messages from 20 WhatsApp groups in the weeks surrounding what many considered to be the beginning of Prime Minister Narendra Modi’s 2024 election campaign: the Ram Temple inauguration in Ayodhya in January 2024. This methodology provides further details regarding the statistical analyses presented in Rest Of World’s investigation “Inside the BJP’s WhatsApp Machine.”

“Ghosts.” “Phantoms.” These are words commonly used to portray the workers training artificial intelligence systems. Despite their crucial role in supporting generative AI, the industry hides their existence. The task of training data systems is far less glamorous than writing code in modern Silicon Valley offices.

Tatiana Dias breaks down what it was like to learn more about these workers, their bosses, and the ultimate beneficiaries of their work.

AI Accountability Fellow Kumar Sambhav Shrivastava and journalist grantee Tapasya investigated opaque welfare algorithms in India that wrongfully cut off benefits to thousands of its poorest citizens. In this piece, Shrivastava shares how the story got started, the questions that drove the reporting, and the lessons they learned.

AI Accountability Fellow Kumar Sambhav Shrivastava and journalist grantee Tapasya investigated opaque welfare algorithms in India that wrongfully cut off benefits to thousands of its poorest citizens. In this piece, Tapasya describes their approach to accessing public records through India’s Right to Information Act, as well as other reporting methods they used to overcome records’ denials and government bureaucracy. 

Lighthouse Reports and WIRED teamed up to examine the growing use and deployment of algorithmic risk assessments in European welfare systems across four axes: people, technology, politics, and business. This methodology explains how they developed a hypothesis and used public records laws to obtain the technical materials necessary to test it.

Lydia Emmanouilidou worked with journalists and researchers to investigate EU-funded high-tech surveillance systems at Greek refugee camps and the Greek border, how they compare to technologies at the U.S.-Mexico border, and U.S.-Greek/European collaboration and lesson-learning on border technology initiatives.

Social Sentinel said its AI technology could help schools prevent suicides and shootings. Our investigation found no evidence that any student lives were saved because of an alert from the service. Our project was a comprehensive examination of the use of social media surveillance software on college campuses. We hope more journalists—particularly student journalists—will continue to examine the impact of artificial intelligence.

On February 24, 2022, Pulitzer Center grantees Karen Hao and Joanne Cavanaugh Simpson joined Reuters Executive Editor Gina Chua for a conversation on AI accountability. The speakers shared how they got started reporting on algorithms, discussed their challenges and breakthroughs, and offered tips for colleagues interested in covering this urgent, underreported story.

Across the country, more than 2,600 police and close to 600 fire departments have partnerships with Ring, the popular doorbell camera company that was acquired by Amazon in 2018. The Markup sought to get a better understanding of what kind of information is sent to police from Ring’s companion app and hyperlocal social platform Neighbors. This article describes our analyses’ data sources, methodologies, findings, and limitations.

A Markup investigation found that Amazon Ring’s social platform, Neighbors, funnels suspicions from residents in whiter and wealthier areas of Los Angeles directly to the police. We teamed up with five students at the Craig Newmark Graduate School of Journalism at CUNY. Together, the students sent public records requests to 25 different police departments. This methodology gives tips for working with students on a public records audit.

Seventy-five percent of the Argentine capital area is under video surveillance, which the government proudly advertises on billboards. But the facial recognition system, part of the city's sprawling surveillance infrastructure, is being criticized after at least 140 other database errors led to police checks or arrests after the system went live in 2019. From the beginning of the investigation, we considered the question of privacy versus security, as well as the regulation of AI and already known racist patterns in facial recognition with the help of AI.

Temperature-detecting cameras. Drones. Technology that police say can predict human feelings. As government agencies quietly deployed new surveillance and predictive tools to monitor their citizens in a time of pandemic and protests, the team at AP compared these tools’ application across nations, and probe how people’s personal data can be sold and mined to expand the knowledge infrastructure of governments and corporations.

What did dating app companies know about the extent of harm on their platforms? When did they know it? What, if anything, were they doing to stop it? Over the course of 18 months, we reviewed hundreds of pages of documents from Match Group, the dating app conglomerate that owns popular apps like Tinder and Hinge, along with thousands of pages of court records, securities filings and analyst reports, coupled with dozens of interviews with current and former employees and survivors of sexual violence.

As part of the investigation, Varsha Bansal conducted a survey of 150 Uber drivers across different parts of India to find out how many of them had been locked out of their accounts—either temporarily or permanently—due to issues related to facial recognition. This investigative effort prompted the gig workers' union to start collecting their own data to petition the platforms.

When trying to understand this content it's common to face a huge volume of it, making any manual approach to understanding this information particularly difficult. 

AI Spotlight Series Coach and AI Fellow Lam Thuy Vo breaks down how machine learning helped her team comb through thousands of videos to isolate trends.

From Freedom of Information requests to using artificial intelligence to analyze satellite imagery, the reporters got their hands on previously unseen data that sheds light on the corruption and systems behind the destruction of the world’s biggest rainforests. Learn about their innovative methodologies.

Using machine learning, The Charlotte Observer and The News & Observer look into a new class of landlords in North Carolina's booming housing market that includes Wall Street hedge funds and other institutional investors. This toolkit is for local and national journalists at a variety of skill levels who are interested in probing the extent of corporate homeownership in their cities, regions, and states.

In the Venezuelan Amazon, traces of the devastation caused by illegal mining can be seen from the sky. This investigation arose from a survey of satellite monitoring information, later processed with artificial intelligence, to see and understand in a comprehensive way the evolution of the mining phenomenon in the Venezuelan Guayana, north of the Amazon.

Mining, one of the main causes of the degradation of rivers and forests in the Amazon, can now be monitored remotely by journalists, scientists, and other concerned citizens. The Pulitzer Center, in partnership with Earthrise Media, has launched the Amazon Mining Watch, a platform powered by an algorithm that analyzes satellite imagery to detect gold mines and other open-pit mining activities in the world's largest rainforest.

In Georgia, a series of Atlanta Journal-Constitution analyses have shown that COVID contributed to hundreds of millions of dollars in increased public debt costs, that Black residents and poorer residents are disproportionately harmed by the bankruptcy system, and that despite all the financial damage that has already occurred, there is a coming wave of bankruptcy filings.

Land banks are vital public agencies who play a key part in turning decrepit, abandoned properties back into viable homes before they attract pests and crime. Using machine learning methods, Eye on Ohio looked at property remediation in several counties to look deeper at a process that has transformed the rust belt over several years.

The true scale of oil well abandonment is likely far greater than the official numbers. In this visually stunning and immersive project, Grist and the Texas Observer modeled oil wells that are likely to be abandoned in the coming years and chronicled the experiences of two Texas ranchers struggling to hold oil companies accountable for polluting their properties.

El Proyecto Dipteryx es una iniciativa periodística basada en el uso de un algoritmo que alerta sobre la sospecha de riesgo de ilegalidad en el comercio de madera amazónica.

Hydropower is a major pillar for sustainable electrical energy generation. The way it is managed, however, is everything but sustainable: Reservoirs and hydropower are a threat to millions, if no measures are taken. Earth observation satellite imagery and neural network-powered coastline detection help to unveil this dark side of hydropower.

Algorithms also have the potential to disproportionately harm some of the most vulnerable members of society by deepening pre-existing social and economic gaps and amplifying racial bias. At the Pulitzer Center we believe this is not just a tech story but an accountability and equity one, too, that should be part of every reporter’s beat.

Pulitzer Center grantee Simpson offered her insights on the topic of artificial intelligence and machine learning (including police surveillance), the intersection of technology and society, and how to approach the tensions between the two in emerging AI technologies as a journalist.

The Pulitzer Center encourages proposals that use advanced data mining techniques, such as machine learning and natural language processing, to solve a data or reporting problem related to a journalistic investigation.

The Pulitzer Center is seeking compelling data-driven storytelling, based on original data collection and analysis and strong visuals, that has the potential to shape public discourse and hold the powerful accountable.