A news website called Inside Arlington has essentially no human content, no garish images, and no catchy headlines.
The website employs artificial intelligence algorithms to report on school committee meetings, property deals, and other local issues in Arlington, a Boston suburb. It was formerly the responsibility of the lowest cub reporters. However, small-town newspapers that formerly offered such coverage are either no longer around or have insufficient manpower to do so.
According to Winston Chen, an Arlington citizen and software engineer, “The town of Arlington is a news desert.” To automate local reporting, he collaborated with fellow Arlingtonian and seasoned overseas journalist David Trilling to form Nano Media, a nonprofit organization. Chen and Trilling intend to sell their technology at a reasonable cost to news-deprived towns across the US if the system succeeds in Arlington.
For anyone to raise their hand and declare, “I want to start this kind of news outlet for my town,” Chen added, “we want the technology to be scalable enough.” This is referred to as local news in a box.
For human journalists who have already witnessed how digital technology has ravaged the business, it is an unsettling concept. According to the Medill School of Journalism at Northwestern University, some 2,500 US newspapers have shut down since 2005, the majority of them in suburbs and small towns. According to the Pew Research Centre, between 2008 and 2020, 57 percent of newspaper writers lost their employment, primarily as a result of newspapers losing billions in advertising revenue to Google and other internet behemoths.
However, it’s not obvious if computer-generated news items will result in a loss of even more newspaper employment. The technology isn’t brand-new, after all.
Cable services For years, Reuters and the Associated Press have produced computerized news, often for routine things like high school and college athletic events or business financial reports. The technology is mostly used by wire services to free up human journalists to tackle topics that are more complicated and demanding.
These programs often operate by consuming raw data from businesses or sports teams, such as box scores or company news releases. They don’t produce the stories using well-known AI tools like ChatGPT. Instead, to make the data easier to understand, these systems mix the stats with language templates prepared in advance by people.
It does not always function. Last month, client newspapers in various towns turned out sports articles that repetitively employed the same twists of phrase, such as “a close encounter of the athletic kind,” making Lede AI, an Ohio business that provides AI tools for creating local sports stories, a national laughingstock. Many of the strange pieces were published by the Gannett newspaper chain, which announced that it had “paused” using the Lede AI system.
Lede AI published a statement that read, “As with any new technological advancement, some glitches can occur.” “We made the necessary changes and immediately began a 24-hour effort to address the issues.”
The technique is advanced further by Inside Arlington. Software that routinely checks governmental websites and downloads the most recent reports is used to find local news. Additionally, it keeps an eye on the city’s YouTube page to find videos of public events. It then creates a written record of each meeting using voice-to-text transcription software. Computers may now cover the school board meeting, negating the need to send a reporter.
Because of the epidemic, many cities now hold public meetings online where computers can listen in, according to Trilling.
Inside Arlington uses ChatGPT software to create the summaries after capturing the messages. The findings are double-checked by humans to remove glaring mistakes. The editor only needs to review it, make sure there are no embarrassing errors, and then click “Publish,” Trilling added.
But are the tales that arise from this journalism? Jeff Jarvis, a former professor of journalism, has his reservations. Jarvis, a co-founder of Entertainment Weekly and author of “The Gutenberg Parenthesis,” cautions against the use of AI in storytelling since it usually creates fictional accounts that have no basis in truth. For instance, the technology news website CNET had to publish revisions to scores of AI-generated pieces early this year because they had glaring factual inaccuracies.
Jarvis cautioned that AI systems had no grasp of reality. “Tying that to an activity where there is an expectation of fact and truth is irresponsible.”
The Initiative for Digital Public Infrastructure director at the University of Massachusetts at Amherst, Ethan Zuckerman, foresees an even more challenging issue. An expert human journalist is knowledgeable about his subject, while AI programs are not. They only combine words using statistics to create stories devoid of context or human intuition.
One of the most challenging tasks we can think of is trying to get meaningful journalism out of this, according to Zuckerman.
For instance, a human reporter would draw attention to anything unexpected or contentious a local official stated during a hearing in his piece. According to Zuckerman, AI wouldn’t notice and would instead write up a dull description of the meeting that would read like “a pile of unseasoned broccoli.”
Unseasoned by incisive analysis, historical context, or even a hint of wit, Inside Arlington’s pieces are little more than a condensed list of statistics. In other words, the website is a deathly, boring read.
Nevertheless, it offers Arlington citizens useful information about municipal administration, which is otherwise quite difficult to get.
Trilling and Chen claim to be cognizant of the limits of AI. They claim that computers are incapable of doing the kind of in-depth research and human interactions needed for serious journalism. Ultimately, Chen declared, “We’re still going to need full-time journalists.”
They might not be nearly as numerous, though.