By Sachi Akmal, CISLM student employee
Generative AI is already in local newsrooms, even as the world debates the best way to use this emerging technology.
On Sept. 25, a group of experts and scholars gathered to discuss how artificial intelligence is reshaping journalism and how it is practiced. Jessica Mahone, research director for the Center for Innovation and Sustainability in Local Media convened and moderated “AI and The First Amendment: How the Press is Harnessing New Technology.” This panel was part of UNC’s annual celebration of First Amendment Day, hosted by the UNC Center for Media Law and Policy.
Mahone led the speakers in conversations on transparency, developing newsroom policies and generative artificial intelligence in reporting and news distribution. Here are four things to consider when using AI for local journalism:
Transparency is key
According to a survey from the Trusting News Project, 94% of readers want journalists to disclose their use of AI. This could look like an editor’s note or byline on articles that explain the use of AI in the reporting process.
The 10th Street Journal, an experimental AI-powered vertical in Duke University’s 9th Street Journal student media outlet, is an example of transparency in action. All stories are published with the tagline, “Articles in The 10th Street Journal are generated by AI and reviewed by human editors.”
Joel Luther, editor of the 10th Street Journal, explained that disclosure was the most important value for them from the beginning.
“The AI product is called the 10th Street Journal, which does not exist, and we’re kind of using that to nod at, ‘Hey, the AI hallucinates, the AI can get things wrong,’” he said. “So we’re poking a little fun at us admitting we know the limitations of AI.”
The experiment was borne out of a lack of student reporters who could cover public affairs news such as road closures or event announcements. The benefit: All these announcements are now available to the Durham community, using AI to fill knowledge gaps in journalism.
Transparency can also look like ensuring your AI and technology policy is visible to your readers by meeting them where they already are, said Mollie Muchna, project manager for Trusting News. She recommends placing your policy on your home page or at the end of articles so that readers can find it easily.
Benjamin Toff, associate professor and director of the Minnesota Journalism Center, echoed that audiences want transparency. “(I think) with some caveat that people will also say that they’re not going to spend a whole lot of time paying close attention to those disclosures, but they do want to be reassured that it is being made available to them,” he said.
Trusting News has a toolkit to help develop your own newsroom policies on AI. These resources do not make recommendations on how to use AI but do encourage transparency and placing policies where people can see them, Muchna said.
It is important to have a set of principles on AI use, but it is also important to consider how they will be policed in the newsroom, said Marisa Porto, UNC Knight Chair in Local News and Sustainability.
Audiences differ, so should your considerations for AI policy
Levels of comfort with AI use are very dependent on the audience, Muchna said. Some audiences are more tech-savvy and comfortable with AI while other audiences may have greater stigmas about its use.
Specific AI tools matter to audiences, too. The same survey from Trusting News found that audiences are more comfortable with journalists using AI to check grammar or transcribe interviews than for translation or data analysis.
Environmental concerns about AI use — specifically, the large amount of power and water required to run generative AI models — may also matter to your readers. Luther cited receiving several such concerns from his Durham-based audience.
AI can help your reporting go further
Human reporting cannot be replaced by AI and will become more valuable, Toff predicted. However, AI can be used to quickly repackage original reporting into formats that can be published across several platforms, such as an Instagram post or TikTok video. This can help reach a fragmented audience with different preferences for consuming news.
From a sustainability standpoint, AI can help local news organizations generate revenue and make better business decisions, said Porto.
News outlets that use a subscription model can use AI to analyze audience preferences and improve readers’ satisfaction with the organization’s content, she said, though privacy concerns must be considered. AI can also analyze business data to inform organizational decisions.
Toff pointed to the Washington Post’s AI chatbot called Climate Answers — which solely pulls from the organization’s reporting to answer users’ questions — as an example of AI helping content go further. He said that this concept addresses two concerns: Protecting the intellectual property of your reporting and increasing online revenue.
Additionally, the prospect that any journalist can do data journalism thanks to AI has tremendous potential, Luther said. Local journalists no longer need computer science skills to sift through large sets of data for their reporting.
Journalists need to be taught AI literacy
Skillful and ethical use of AI in journalism begins in the classroom. Porto believes journalism curriculums should include the history of AI and generative AI, the differences between these technologies and the ethics behind their development and use.
Additionally, journalists should have a level of media literacy that helps them discern AI-generated content. For example, journalists need tools to help them verify whether the voice on the other end of the phone is AI-generated or the intended interview subject, Toff said.
The panel expressed a range of concerns for the use of AI if it is not intentionally regulated. As Porto put it, technology requires humans to drive it. She is confident that journalists will help develop the guardrails necessary for the sustainable integration of AI in the field.