IIt didn’t take long for the shortcomings of the AI-powered generative news editorial office to become apparent. CNet’s The secret chatbot reviewed editors last November, of buzzfeed Human staff were subsequently laid off in favor of AI-generated “content” creators. Fear of being replaced by “sufficiently good AI” has been lingering in the minds of many journalists lately, with one-third of the country’s newsrooms expected to close by the mid-2010s. It is
But AI need not necessarily be an existential threat to the field. As six research teams demonstrated at the New York University Media Lab’s AI & Local News Initiative demo day at the end of June, the technology could be key to fundamentally transforming how local news is collected and produced.
Now in its second year, the initiative has a mission to help local news outlets “leverage the power of artificial intelligence to drive their success.” It is supported as part of a larger $3 million grant from the Knight Foundation, which Associated Pressthe Brown Institute’s Local News Lab, the New York City Media Lab, and Partnerships on AI.
This year’s cohort includes a mixed team of academia and private companies who, through a 12-week development course, will “empower journalists, support sustainability in the press, and provide quality information to local communities.” We have united to build an AI application for local news. News Viewers,” reported New York University’s Tandon News Agency.
“There is value in being able to bring people together who are working on these issues from different angles,” Matt McVay, community and project lead for the effort, told Engadget. ”
“This is also an opportunity, because if news outlets that are doing a good job can continue to communicate their values and maintain the trust of their readers,” he continued. “I believe that as reliable news sources become more readily available, we will have an information ecosystem.” [for anyone] lower quality [AI generated] content. “
The six teams include Bangla AI, which is developing a web platform that displays and translates relevant news articles into Bengali for journalists and New York City’s large Bangladeshi immigrant community.
“More than 200,000 Bangladeshi legal immigrants live in the United States, half of them in New York City,” Dr. Ashraful Goni, a member of the Bangladeshi team, told reporters on the day of the demonstration. “Only half the population speaks English fluently,” and the other half lacks easy access to the day’s news through mainstream media such as television and newspapers. new york times or Associated Press.
“Bangla AI retrieves and translates information that is relevant to people in the Bengali community published in mainstream media. The system will also generate summaries of mainstream media posts in both English and Bengali, allowing local journalists to cover important news rather than rewriting telegraph copy.
Similarly, the team at Chequeado, a non-profit that fights public misinformation, showed off the latest development of its Chequeabot platform, Monitorio. Leverage AI and natural language processing capabilities to streamline fact-checking efforts in Spanish-language media. Its dashboard continuously monitors social media for trending misinformation and can alert fact-checkers to slow the spread of your work.
“One of the biggest hopes for this stuff and Bangla AI is,” Chequeado team member Marcos Barroso said during the demo. “It’s about being able to bring this kind of technology into underresourced newsrooms to improve their capacity and improve their ability to be more efficient.” ”
The Newsroom AI team at Cornell University hopes its writing assistant platform will do the same for journalists as Copilot did for programmers: remove the tedium. Newsroom can automate many common tasks, such as transcription and organizing information, image and headline generation, and SEO implementation. The system can also reportedly write stories in a journalist’s personal style, given enough training examples.
On the audio side, the team at New York Public Radio WNYC spent time developing and prototyping a speech-to-text model that produces real-time captions and transcripts for live broadcasts. WNYC is New York’s largest public media station, with 2 million monthly visitors through its news website.
“At this time, our live broadcasts do not have a meaningful entrance for deaf and hard of hearing viewers,” WNYC team member Sam Guzik said during the demo. “So what we really want to think about as we look to the future is, ‘How can we make our audio more accessible to people who are deaf?’”
Using AI to perform speech-to-text conversion alleviates one of the biggest bottlenecks in modern subtitling. The point is that it takes a lot of money and resources to do the work quickly if you let people do it. “The speech-to-text model is relatively low-cost,” he continued Guzik. “They can operate at scale and support an API-driven architecture that ties into our experience.”
The result is a proof-of-concept audio player for the WNYC website that produces accurate closed captions for the currently playing clip. The system can go one step further by summarizing the contents of that clip into a few bullet points at the click of a button in the audio player.
“This is a meaningful improvement, even for deaf people,” Guznik said. “But even for people who aren’t in a listening space, it’s a great tool if you want to follow what’s being said where you don’t have headphones.”
On the backend, NOBL Media has developed an ad tech product that “enables programmatic advertisers to reach content from publishers serving small audiences that can be targeted based on geography and demographics.” Meanwhile, Graham Media Group created automated natural language text. Encourages the comments section of local news articles to move in the direction of civility.
“Comment bots can post the first comment on a story to guide the conversation and hopefully increase participation to take users deeper into the engagement funnel,” GMG team member Dustin Block said during the demo. Stated. This solves his two key challenges facing human comment moderation. It is to prevent the loudest voice from dominating the discussion and to provide form and structure to the conversation.
The bot uses the GPT 3.5 Turbo API to scan and understand news articles. It creates thought-provoking cues and encourages subsequent discussion,” he continued. “Made to be friendly.”
It remains to be seen whether the AI revolution will remain journalist-friendly, and perhaps even grow, but McVay isn’t worried. “Most news outlets, especially local ones, are so under-resourced and staffed that there’s more going on than they can cover,” he said. “So, AI and [the automations seen during the demo day] Increase bandwidth for journalists and editorial staff. ”
All products recommended by Engadget are selected by an editorial team independent of the parent company. Some articles contain affiliate links. When you purchase something through one of these links, we may earn affiliate commissions. All prices are correct at time of publication.
