Nieman Foundation at Harvard
HOME
          
LATEST STORY
ProPublica’s new “50 states” commitment builds on a decade-plus of local news partnerships
ABOUT                    SUBSCRIBE
May 2, 2024, 2:56 p.m.
Reporting & Production
AI Spotlight Series

Pulitzer’s AI Spotlight Series will train 1,000 journalists on AI accountability reporting

The Pulitzer Center is prioritizing reporters in the Global South, and all the sessions are free.

The Pulitzer Center has officially kicked off The AI Spotlight Series, a new training initiative that aims to teach 1,000 journalists how to do AI accountability reporting over the next two years.

On April 21, roughly 40 journalists gathered at the University of California, Berkeley for the inaugural “Introduction to AI reporting” session, which was built to demystify basic AI concepts for reporters outside the tech beat.

“I had this hypothesis that if journalists could understand the fundamentals, they would naturally start making connections and would start thinking of more story ideas,” said Karen Hao, the lead designer of the series and a contributing writer for The Atlantic, who spent years covering the rise of AI at MIT Technology Review and The Wall Street Journal. “I was watching them do that in real time.”

The AI Spotlight Series stands out for more than its sheer scale. The Pulitzer Center is explicitly prioritizing journalists outside of North America and Western Europe in putting the training sessions together.

“So much AI coverage is done by the Global North focused on the Global North. To me, we’re just missing such a huge part of the story,” said Hao, who reported a Pulitzer Center-backed series on AI colonialism during her time at Tech Review. “If we want a technology that is so consequential to actually benefit all of humanity — as OpenAI likes to say — the best way to understand how to do that is by covering the communities that are the most vulnerable and have had the least amount of agency in shaping the technologies thus far.”

While some of the biggest AI developers in the world may be headquartered in Silicon Valley, the supply chain that powers their products extends far beyond California. That includes data workers in Venezuela, Pakistan, and the Philippines who help produce the training sets that are fed into AI models, and the content moderators in Kenya who refine tools built with them.

Increasingly, reporting has also shown it is outsourced and gig workers in the Global South who feel the brunt of early adoption when these AI technologies enter the workforce, whether through job displacement, worker surveillance, or algorithmic management.

“We must take a very global approach to doing this coverage, with these trainings, and this literacy work, otherwise we’re going to perpetuate the same approach that the AI industry has taken,” said Marina Walker Guevara, executive editor of the Pulitzer Center. “It’s our challenge.”

The AI Spotlight Series will run simultaneously on three tracks, with each type of training session targeting a different category of journalist. The first track is focused mainly on reporters who aren’t on tech desks and who may not have any previous exposure to AI basics.

“When it comes to technology reporting, generally, there is a lot of jargon-based gatekeeping,” said Lam Thuy Vo, an investigative reporter at The Markup and one of the series co-designers. “People are throwing around big terms, like deep learning, neural networks, machine learning, artificial intelligence, and so on.” Vo says this first layer of training — effectively AI reporting 101 — will help reporters understand “the basic plumbing of AI.” The goal is to make them confident enough to investigate the ways these technologies touch down in their respective beats and in the communities they cover.

Imagining potential participants for the first track, Hao suggested a breaking news reporter covering Elon Musk’s latest tweet, a health desk reporter tracking how AI is entering a hospital system, or a political journalist anxiously covering deep fakes ahead of an election day. “If we did not target them, we would be missing a huge part of the coverage that is shaping the public discourse,” she said.

The second track is application-based and will pull from a pool of reporters who are already well-versed in AI coverage. These longer, in-depth sessions are currently capped at 25 to 30 journalists and will train them to conduct more technical investigations into topics like algorithmic bias.

Top of mind for Hao and Lam is pushing these reporters to avoid common tropes and pitfalls in AI beat reporting. One goal for their coverage is to stop flattening the complex web of technologies known as AI into just ChatGPT or Dall-E. Hao points out that many newsworthy AI technologies impacting healthcare, policing, and climate change have little to nothing to do with generative AI.

The final track will target editors, whom Walker calls the “gatekeepers” to AI accountability coverage.

The Spotlight Series is housed under the Pulitzer Center’s AI Accountability Network, a program that launched in 2022 to support the work of journalists investigating predictive and surveillance tech around the world. Walker says historically many reporters in the program have a difficult time placing their stories with editors at more mainstream publications, including local news outlets.

Common responses from editors to AI-related pitches, according to Walker, include “This sounds fascinating but we don’t cover tech” or “Let me introduce you to my friend at Wired.” Many reporters in the network ultimately run their pieces at dedicated tech publications like MIT Technology Review, The Verge, or Rest of World. “I was like, no, no, this is about your city, this is happening in your backyard, this should interest you,” she said.

Walker hopes these targeted sessions, which are also capped and require an application, will bring more editors into the fold of AI accountability work. The sessions are designed to walk editors through story framings that aren’t exclusively about “tech,” but also about labor, politics, immigration, policy, education, or criminal justice.

“I wouldn’t want us to successfully train up all the reporters, and then they come into their newsroom and have to do a lot of work to get the editors on board,” said Hao. For better or worse, an editor’s subject knowledge does shape coverage and resource allocation. “We want to help them make those high-level strategic decisions of where to put their resources. Which stories do they put multiple reporters or the graphics team on? How do they make the big investigation or feature bet?”

While The AI Spotlight Series welcomes any journalist to join or apply, the registration page explicitly states it will prioritize those “from the Global South and from communities underrepresented in media.” That sentiment is more than lip service. In-person workshops have already been scheduled for the International Media Conference in Manila, Philippines in June and the Gabo festival in Bogotá, Colombia in July. The Pulitzer Center is also pitching the African Investigative Journalism Conference in South Africa in November and the Arab Reporters for Investigative Journalism annual forum in Jordan in December.

Between these sessions, the Series will be hosting virtual workshops on Zoom in several time zones to make it easier for journalists across Asia to attend, as opposed to scheduling around Western journalistic hubs like New York and London.

The initial slate of sessions are in English, but the Pulitzer Center also has plans to roll out simultaneous live translation in French, Spanish, Portuguese and Bahasa Indonesia. Walker noted they’ve already received requests to add additional languages, including Arabic. The first non-English sessions will begin this fall.

Most notably, all sessions in the AI Spotlight Series are free to register for each of the 1,000-plus participants they are targeting over the next two years. Alongside the Pulitzer Center, the series also has backing from the MacArthur Foundation, Ford Foundation, and Notre Dame-IBM Technology Ethics Lab.

Hao said all the co-designers had agreed upfront to take a pay cut if the additional funding didn’t come through. “We didn’t want to jeopardize the sessions not being free, because that would defeat the purpose,” she said. “We do not want this to be exclusive knowledge in any way.”

Andrew Deck is a generative AI staff writer at Nieman Lab. Have tips about how AI is being used in your newsroom? You can reach Andrew via email (andrew_deck@harvard.edu), Twitter (@decka227), or Signal (+1 203-841-6241).
POSTED     May 2, 2024, 2:56 p.m.
SEE MORE ON Reporting & Production
Show tags
 
Join the 60,000 who get the freshest future-of-journalism news in our daily email.
ProPublica’s new “50 states” commitment builds on a decade-plus of local news partnerships
With annual revenue of $45 million and a staff approaching 200 people, ProPublica has been one of the big journalism winners of the past decade. And it’s been unusually willing to spread that wealth around the country.
“Journalism moves fast…philanthropy moves slow.” Press Forward’s director wants to bring them together
“I see, every week, some example of where the two don’t understand each other. Each of them needs to shift a little bit.”
After criticism over “viewpoint diversity,” NPR adds new layers of editorial oversight
“We will all have to adjust to a new workflow. If it is a bottleneck, it will be a failure.”