By Ian Dartley
It all started with a dog.
“I was walking my dog around the dog park one morning,” said Winston Chen, a resident of Arlington, Massachusetts. “During the pandemic, leash laws weren’t enforced. Now they’ve started getting fined for having dogs off leash. All the dog owners were upset.”
But when Chen recommended they approach the town’s parks and recreation commission, he received groans.
“Two and a half hours? They had no time for that,” Chen recalled. So he teamed up with his neighbor, longtime journalist David Trilling, to brainstorm ideas on how they could empower artificial intelligence to deliver the local government’s services to the people of Arlington. They started a nonprofit, Nano Media, which uses AI to power their first local news project, Inside Arlington.
In brief
-
- Arlington residents use AI-driven nonprofit, Nano Media, to simplify access to local government information through Inside Arlington.
- The organization automates content, particularly through transcription of government meetings, to enhance information accessibility.
- Nano Media aims to empower local news with AI, emphasizing its role as a tool to aid journalists and engage citizens in the community.
Trilling and Chen recently discussed their work at Northeastern University’s School of Journalism. They told students and faculty members they had identified a gap in awareness of the town’s inner workings. Inside Arlington, they said, can distill information residents need in a timely and organized fashion.
They coined the idea “local news in a box.” With the exception of the police log (privacy concerns complicate that), they wanted to automate as much content as possible. One way was through the transcription of government meetings.
They train their model to learn names and exercise judgment by feeding it transcriptions of board meetings. As a result, Chen said, these transcriptions are “almost as good as a human’s.” He added: “Some of the tools and processes we use are proprietary. But we use a variety of AI-based tools, such as large language models,” known in the AI world as LLMs.
Trilling said that Inside Arlington values accuracy over flash. “We wanted to make it as boring as possible,” he said. “When the AI has to exercise judgment, it often does a decent job. But it doesn’t have the context to make a good judgment. Of five topics discussed in a meeting, which is most important to a reader? It can’t distinguish that. It doesn’t know if something has been a major topic in a town for a long time.”
As AI continues to careen into the path of journalism, most obviously through Gannett and the MSN news service’s recent controversies where several sports articles were completely botched (in the MSN article, a deceased NBA player was called “useless” and in Gannett’s case, stories about high school football games were littered with errors and strange language), there were understandably some concerns about what sites like Inside Arlington might mean for local news.
Did Arlington need another news site? The town already has a nonprofit news organization, YourArlington, doing things the old-fashioned way — with human reporters. Some of those in attendance at Northeastern were skeptical.
“I think a lot of us look at local news as part of an ongoing conversation a community is having among themselves,” said a faculty member. “The process is as important as the result, a way of building civic engagement. How does increased automation contribute to civic engagement? You can make an argument it detracts.”
Chen responded that AI doesn’t eliminate the human dimension. “We do not advocate for a world where AI drives the flow of information,” he said. “We believe that humans belong in the driver’s seat. Whether it’s AI-driven or something else, it’s helping the information and making it more digestible.”
Chen and Trilling said they believe that converting a three-hour video into an easy-to-read summary of a zoning meeting drives civic engagement by making that content more accessible. Their prime example of this is a feature on the website that allows residents to find out easily about local ordinances.
Chen demonstrated by navigating to a part of the Inside Arlington website that had a chatbot. He typed in a question about how a resident can obtain a permit for raising chickens and hens, and the chatbot spit out a full, detailed response within seconds.
While this tool wowed the audience, there were still further concerns about the ongoing integration of AI and journalism. One student asked how they could reassure journalists that their technology is a tool for journalists to use, not a replacement that will leave them unemployed.
“I hope journalists have an open mind,” Trilling answered. “They’re being replaced by the Gannetts of the world, not AI.” Trilling stressed that their site is a tool that can help journalists thrive and outlets flourish, adding, “It shouldn’t replace humans. It should make them more productive.”
The rest of the discussion centered on how their site will help residents of Arlington communicate with town officials. Toward the end, one student asked whether their tool may end up doing more harm to local news in the long run than good. Trilling and Chen answered by saying they believe that their AI tool addresses a problem instead of creating a new one.
“Local news has so many challenges,” Trilling said. “You’ve got to do something to help it.” The Nano Media website says that the nonprofit’s mission is “to foster an informed and engaged citizenry by revitalizing local news in an economically sustainable way.”
But though Trilling and Chen’s creation might be helpful in communities with no local news, it’s certainly within the realm of possibility that other newsrooms, especially those owned by cost-cutting corporate chains, will see this as little more than an opportunity to save money.
The future of their project is not clear, but their endgame, they said, is to find a way to streamline local news about issues ranging from taxes to development. And they stressed that they see Inside Arlington as an experiment.
“We’re throwing spaghetti at the wall and seeing what sticks,” Trilling said.
Editor’s note: In keeping with the theme of this story, we asked ChatGPT to write the headline and the bullet points summarizing the article. We also used ChatGPT to write social-media posts promoting the article.
Ian Dartley is a graduate student in Northeastern University’s School of Journalism.
What Works 