How Journalists Can Use AI Technologies to ‘Speak Truth to Power’

Aimee Rinehart was working as the program manager for The Associated Press’ (AP) Local News Artificial Intelligence (AI) Initiative when OpenAI launched ChatGPT on Nov. 30, 2022. It didn’t take long for the AP’s senior product manager of AI strategy to realize the chatbot that she and millions of others were experimenting with would revolutionize the journalism industry. 

“We were ready to meet the moment,” Rinehart said. “It’s an exciting place to be.” 

Rinehart was a guest speaker on Jan. 30 for the AI@NU Graduate Student Group’s “How AI Is Transforming Journalism” event. She discussed how local newsrooms can incorporate AI into their work and addressed concerns about the technologies, such as bias and job displacement. Another fear, the possibility that journalists’ work will be devalued, has already disrupted many newsrooms amid the rise of online platforms and social media. 

The AP Local News AI Initiative began in August 2021 through a grant from the John S. and James L. Knight Foundation. This was before the hype surrounding generative AI platforms, during a time when people weren’t fully convinced whether AI was going to be the next big thing.   

“At the time, no one really cared,” Rinehart said. “It sounded a little like virtual reality headsets, cryptocurrency, blockchain, or somewhere in between. All of that stuff hadn't really taken off. It was something they’d roll their eyes about because people weren’t able to monetize the technologies.” 

After ChatGPT was released to the public, Rinehart realized that she didn’t need to make many sales pitches about the benefits of AI because local news leaders understood that the technologies were going to disrupt the industry. While some were worried about the potential changes, others were excited about the opportunities. 

“We have not reached equilibrium since the mid 1990’s in terms of innovation slowing things down,” Rinehart said. “Those who resist the change get left behind. They don’t get promotions. They’re not included in things. I like to be included. I like to be out in the front.” 

Since AI is everywhere, Rinehart told aspiring journalists to understand when governments may be using the technologies. For example, in 2020, Tampa Bay Times reporters uncovered that the sheriff in Pasco County, Fla. was using school district data about grades and attendance, as well as state Department of Children and Families data about abuse histories to have algorithms predict whether schoolchildren were likely to become criminals. Their investigation won a Pulitzer Prize.     

“As a journalist, if you don’t know the tools that power is using, you can’t speak truth to power,” Rinehart said. 

Through nearly 200 surveys, Rinehart and the AP learned that local news leaders felt AI could help journalists. AI technologies can transcribe interviews, generate story ideas, and suggest headlines, among other tasks. However, news leaders were also concerned about the costs of using AI and the extensive training process.  

Rinehart and the AP also spoke with local news leaders who are worried that consumers will get their information from large language models (LLMs) instead of traditional news sources. Companies like Google and Microsoft have incorporated generative AI into their search engines. Rinehart said that while it’s unlikely that LLMs will link back to a local news site’s coverage, there are steps that newsrooms can take to combat the decrease in referrals. They can focus on platforms they own and operate, such as an app, podcast, or newsletter. 

“Having your own app or newsletter isn't going to make it all go away,” Rinehart said, “but it could stop some of the severe bleeding.” 

The open-source community has resources that local news leaders can use to build large language models. Rinehart believes this is the moment for the journalism industry to have “radical collaboration” to support its own LLM for news. 

“We’ve been on the back foot with big tech for the past 20 years,” she said. “We did not make great decisions when it came to the web. I just hope that through all of that painful learning, we can come together and do this. We’re really great at information, and I think we should be great at distributing it as well.” 

OpenAI has examples of tools that Rinehart believes the journalism industry can mirror to automate certain tasks, such as updating community calendars and subscription lists. She said the industry also needs people with computing skills like prompt engineering. 

“Most journalists want to be constant learners,” Rinehart said. “We’re better when we think through these things together.” 

The AI@NU Graduate Student Group regularly organizes events featuring speakers who explore the use of AI across different industries, including talks by renowned scientists in AI and panel discussions with experts in the field. To learn about more AI-related events at Northwestern, visit the events page on the AI@NU website.  

Members of the Northwestern community can watch the full recording of the event on Panopto.