AI-generated news has arrived in New Jersey
It’s not journalism, but does it have a role to play in the information ecosystem?
Imagine you’re using a metal detector to search for buried treasure at the Jersey Shore. You hear the metal detector beep and you start digging, eventually unearthing an old coin from deep in the sand.
Of course, you can’t ask the metal detector to tell you the story of how this particular coin ended up buried on the shores of Asbury Park, but it did do its job of alerting you to its presence, encouraging you to take a closer look.
Now I want you to imagine that I wrote a really smooth transition sentence here, and consider this: A new, completely AI-driven website called the LocalLens wants to be a kind of metal detector for local news — claiming to surface stories that might otherwise remain buried.
The first questions that likely come to mind when journalists and publishers hear about this are: Can we really rely on machines to produce journalism and news stories that matter? Should we? Can AI really capture the nuances of human life and local color that form the backbone of local news? Is this really the future of local journalism?
The short answer is, of course, no.
The longer answer is also no, but that doesn’t mean these tools can’t be useful for newsrooms who know how to use them without losing their connection with the communities they serve.
Generative AI has already started creeping into the journalism landscape, with promises that it will allow journalists to quickly sift through vast amounts of data and churn out content at a pace that no human journalists could hope to match. But while the scale and speed are certainly impressive (I personally use bots like ChatGPT all the time, so much so that I wrote a whole handbook on how local publishers and journalists can use these new tools), concerns about accuracy, ethics, and the loss of the human touch in storytelling are growing.
LocalLens, launched by Allendale School Board member Mat Hernandez, claims to use Large Language Model (LLM) technology and government records to automatically generate coverage of local government. The site says the point of LocalLens is to increase the visibility of local government activities, serving as a starting point for journalists.
The “Our Mission” page says the goal of LocalLens is to “democratize local democracy” by “leveraging new technologies to make the work of local government accessible to every American in every community.” It also says the ultimate goal of LocalLens is to “cover what’s happening in every community in America” and to “be helpful not only for knowing what’s happening in your own town, but also in communities across the country.”
My immediate reaction to that statement is: Lol, okay. Sure.
Further down the page, they make sure to clarify that this kind of bot-driven coverage is “no replacement for in-depth local reporting.” Instead, LocalLens aims “to be part of the infrastructure that makes local news possible by serving as a resource for local journalists stretched too thin to attend every local meeting.”
Fair enough, especially when you look back on the last few decades of devastation via layoffs, buyouts, closures, and hedge fund-ifications impacting virtually every local news organization across the country.
Record number of media job cuts so far in 2023
The level of cuts is worse than at the outset of the pandemic in 2020.
But if we pause for a second and give Hernandez and LocalLens the benefit of the doubt, the question becomes: With AI at the helm, how accurate is this visibility? And how much depth is lost in this robotic process?
At first glance, it doesn’t appear that Hernandez or anyone else at LocalLens takes time to review or edit the copy generated by their news bots. In fact, the site even has a little disclaimer at the bottom of the page that warns:
Every story you read on the LocalLens is written by a computer. There may be inaccurate information about people, places, or facts. If you spot a mistake, please let us know and we will fix it. You should review the source link at the bottom of every article before taking action based on something you read here.
The only clue I could find that suggests a human might be at least somewhat involved in the process is a typo in one of the recent “stories” published on the site.
I spoke to LocalLens cofounder, Matt Post, on a Zoom call for a group of local news publishers interested in the relationship between local news and generative AI, and he confirmed that they do not in fact review or edit the content that his bots generate.
He cited one recent example, in which the name of a school superintendent was misspelled, and the superintendent emailed LocalLens asking for a fix. But other than that, Post says the copy is the copy.
“We don’t edit the stories at all after the models spit them out,” Post said. “But if there’s something that’s really bad, we go in and fix it.”
A recent example of LocalLens’ AI “reporting” was picked up by Matt Friedman in his New Jersey Playbook newsletter for Politico.
The LocalLens bot appears to have scraped the livestream transcript of a New Jersey school board meeting and generated a “story” titled, “Valedictorian’s Candid Speech on Racial Discrimination Prompts Censorship Push by School Board.”
The copy isn’t terrible. It reads relatively well, it conforms to the basic “inverted pyramid” style of straight news reporting, and it highlights key points from the school board meeting.
It even includes thoughtful musings like this one on the inherent irony of the school board’s attempt to prevent the student from giving her speech:
One could argue that the school board’s response, rather than addressing the issues of racism and stereotyping that Shah bravely brought to light, is a testament to the urgency of her message.
Not bad, but it also seems to be dipping its generative toes into an area of analysis and commentary that goes slightly beyond LocalLens’ purported mission to simply serve as a “resource for local journalists stretched too thin to attend every local meeting.”
A bit nitpicky, perhaps, but I don’t think it’s a stretch to assume that the average reader might not notice, understand, or even care about the nuances and distinctions of “actual journalism” vs. bot-generated summaries and breakdowns.
Which makes me wonder what LocalLens might look like in the near future once they’ve expanded to “cover what’s happening in every community in America.” Do the founders of LocalLens see themselves as something akin to a bot-driven version of Documenters that supplies much-needed notetaking and documentation of public meetings and records? Or will this eventually just become yet another “good enough” source of local news and information for residents in communities that have either lost or abandoned their own local news organizations?
I’m sure for most local journalists and publishers in New Jersey and elsewhere, the answers are obvious — and probably somewhat troubling.
A few other questions that come to mind:
- What kind of prompts does LocalLens use to guide and instruct these bots as they scrape, analyze, and generate summaries of these documents and meeting transcripts? (Matt Post says he’s open to the idea of sharing his prompts.)
- How does LocalLens pay the hosting and LLM costs associated with running the site and generating the content? (The site says LocalLens is “a fiscally sponsored project of Ready For School Initiative, Inc., a 501c(3) nonprofit organization,” but that’s it. There are also no human beings listed under the “About Us” page and the “Contact Us” page is just a basic form embed.)
- Will they eventually start running ads on the site? What about sponsored content?
- Will they accept submissions from outside orgs, individuals, or political parties for their bots to process and turn into content?
Ultimately, AI is a prism. Information goes into it and the bot can refract a spectrum of stories and simulated perspectives, but it may also distort those views, missing the nuances and human elements that give local news its heart and soul. Understanding and staying abreast of these technological developments is crucial, but so is maintaining a healthy skepticism.
As these bots and tools continue to work their way into the field of journalism, I believe it’s important for their role to remain assistive, not dominant and unchecked.
Just as a metal detector can’t tell you the story behind a buried coin, these bots can’t provide the depth and nuance that comes from human reporting. And as we navigate this new landscape, it’s crucial to remember that the charm and value of local news and community information comes from the human, not the machine.
About the Center for Cooperative Media: The Center is a primarily grant-funded program of the School of Communication and Media at Montclair State University. Its mission is to grow and strengthen local journalism, and in doing so serve New Jersey residents. The Center is supported with operational and project funding from Montclair State University, the Geraldine R. Dodge Foundation, Democracy Fund, NJ Civic Information Consortium, Rita Allen Foundation, Inasmuch Foundation and the Independence Public Media Foundation. For more information, visit centerforcooperativemedia.org.