When Tomás Dodds set up the Public Tech Media Lab, he wanted to create a place where research could go beyond the university and into local newsrooms.
This year, Dodds collaborated with Isthmus, an independent, local news source based in Madison, to develop guidelines for using artificial intelligence in the newsroom. This initiative aligns with his broader interest in helping journalists navigate new technologies without losing the values and practices that underpin journalism.
“When I created the Public Tech Media Lab, we knew that we didn’t want this to be an academic silo, kind of this idea of the academic ivory tower, where we just write and think for other academics,” Dodds said. “I knew that I wanted to work with journalists and practitioners and with the industry in solving ongoing problems.”
To Dodds, that meant starting with local newsrooms themselves.
“Wisconsin has a rich ecosystem of media organizations, but they’re facing the same challenges that AI systems are presenting to newsrooms all over the world,” Dodds said. “Whether it’s a public broadcaster or a small local paper, everyone is trying to figure out how to use these tools responsibly, how to keep their audience’s trust, and how to make sure the technology serves journalism. It’s not an easy job.”
For Isthmus and Dodds, the work began with conversations about how employees were using AI and how the newsroom planned to use it going forward.
As Dodds puts it, drawing up guidelines is not just about informing journalists what they can and cannot do with AI. It was also about establishing a common understanding among newsroom members on how the technology should be used.
Transparency and human-in-the-loop were the two main principles of that approach.
According to Dodds, transparency involves informing readers about journalists’ use of AI, including in cases when things go wrong. By remaining open, guidelines can help news organizations in keeping the trust of their audiences.
“What being transparent allows you to do is this: if you ever make a mistake, people are more willing to trust you again if you can point to guidelines and say, ‘Hey, look, we made a mistake. That is what went wrong, but we are not hiding anything. This is how we use AI,’” Dodds said.
The guidelines also highlight the principle that journalists should stay involved throughout the process. Dodds calls this keeping a “human in the loop,” meaning journalists need to oversee how AI is used instead of letting it make decisions alone.
“You need to tell people how AI is being supervised by different humans in different stages of the process,” Dodds said.
That approach gives the guidelines a purpose beyond simply telling journalists what they can and cannot do. The guidelines set out what the newsroom hopes to accomplish with AI while also giving journalists a shared understanding of how to use the technology.
The guidelines cannot, however, stay fixed. As AI tools keep developing, newsrooms should review their policies and adjust them as needed. Dodds referred to the guidelines as ‘living documents’ which should be revisited from time to time. Emphasizing how this should be the case for all guidelines, even before the use of artificial intelligence.
Rather than drawing up policies based on particular tools which may soon become obsolete, Dodds hopes that newsrooms concentrate on principles which can develop over time. He also sees preparing AI policies as an opportunity for journalists to review their own work.
As Dodds points out, the problem with the conversation about artificial intelligence in the newsroom is that “it seems that artificial intelligence is dangerous and something we need to be extremely careful about.”
While this is true, Dodds also sees AI as an opportunity to help journalists audit their work.
“Artificial intelligence is a huge opportunity and allows you to do different things, either for research or engagement with the audience, and so we want to develop these small AI tools for local newsrooms and help them use AI in a way that is ethical,’’ Dodds said.
Dodds’ AI guideline work with Isthmus is an incredible opportunity. Instead of avoiding the conversation about how AI will affect journalism and dealing with it later, the guidelines give newsrooms a way to decide how to incorporate AI while keeping journalistic values at their heart.
The lab is open to working with other news organizations that want to create AI guidelines. Newsrooms interested in learning how to use AI in their work can reach out to the lab using this email: ptml@sjmc.wisc.edu