Innovation

The Beat Book will employ artificial intelligence to help local news organizations assess and improve coverage of their communities. Using large language models and other AI methodologies, the Beat Book will scour a cooperating publication’s archives to identify how that publication has covered a particular beat in the past.

The book will assess not just what the publication had done but also what it had not; which community members, authorities and experts it had talked to and reported on, and who it had not; and whether coverage of a neighborhood had involved a range of frames and topics or portrayed it through a narrower, less complete lens.

Merrill College faculty member Derek Willis is already working on the book.

The AI Reporter’s Tool Box focuses on how to harness AI to improve journalism and provide value to consumers. An idea initially developed by students in a Merrill College class, the tool box will use AI and other machine learning technology to make the reporters’ tasks more efficient and their reporting more accurate.

That will begin with the most basic element of journalism — the interview. Synthesizing information from interviews has always been time-consuming. With automated transcription services, reporters no longer need to manually transcribe their work, but combing through long transcripts can take even more time than handwritten notes once did.

Large language model AI can make that process vastly more efficient: instantly summarizing key points; extracting key quotes that explain those points; producing story-quality paragraphs that a reporter could quickly include in a story; suggesting follow-up questions; linking to contextual, external resources or prior coverage related to a specific statement; and fact-checking key statements. Faculty members Sean Mussenden, Tom Rosenstiel and Derek Willis co-lead this initiative.

Journalists covering local communities may spend a lot of time sitting in long government meetings. They often contain news but can be an inefficient use of reporters’ time in shrinking newsrooms. This AI system would address that problem by monitoring live-streaming video of government meetings, and providing timely notifications of newsworthy events, story idea tips and summaries of key action.

Such a system would be trained on years of prior government meeting video and transcripts, historical news coverage, a news organization’s coverage priorities, meeting agendas and other inputs.

The goal is not to produce automated stories but to provide a tipsheet that could free up time for a reporter to produce higher-value coverage. Faculty member Sean Mussenden is leading research and development on this system.

Search engines and social media are two of the main avenues for web traffic for publishers. Yet algorithm audits of these platforms show local news sites are underrepresented in search results and social media feeds. Previous research shows these platforms look for specific markers of quality, relevance and authoritativeness in websites in order to give them priority.

The Visibility for Local News project will work to determine which of those markers have the most impact and whether local news organizations are likely to use them.

Led by faculty member Daniel Trielli, the project is a multiyear initiative combining research, teaching and partnerships with local news organizations. The insights from this research will be used in courses for Merrill undergraduate students, who will collaborate with Maryland local news to help them improve digital visibility and track changes in audience.

The Howard Center for Investigative Journalism, launched in 2019 and funded by the Scripps Howard Foundation, gives Merrill College students the opportunity to work with news organizations across the country to report stories of national or international importance to the public, guided by college faculty and staff.

The multidisciplinary program is focused on training the next generation of reporters through hands-on investigative journalism projects. Students learn to dive deep into data, and tell the stories they uncover in new and compelling ways. The UMD Howard Center has won more than a dozen of the nation's most prestigious professional journalism awards, and been a finalist for several others — primarily for innovation and collaboration. 

Howard Center students: 

• Built temperature and humidity sensors to help understand what it was like to live in Baltimore’s hottest neighborhoods, and show the disparate effects of climate change. 
• Conducted a first-of-its-kind analysis of mobile-phone location data to prove the majority of customers at lottery retailers come from nearby neighborhoods and, using census data, that those neighborhoods are disproportionately home to Black, Hispanic and lower-income people. 
• Used AI to create an unprecedented database of House travel, finding and mining documents that had been removed from public view.

The center has collaborated on its investigations with a variety of professional and academic partners, including The Associated Press, NPR, PBS NewsHour, FRONTLINE (PBS), USA TODAY and universities across the country.