AI's impact on local news

Medill's Jeremy Gilbert addresses the Mega-Conference

Jeremy Gilbert, professor and Knight Chair in Digital Media Strategy at Medill, Northwestern University
Jeremy Gilbert, professor and Knight Chair in Digital Media Strategy at Medill, Northwestern University

AI is impacting the local news environment at a rapid speed.  During a talk at the 2024 Mega-Conference, Jeremy Gilbert, professor and Knight Chair in Digital Media Strategy at Medill, Northwestern University, highlighted the potential for AI to transform the news industry, enhance performance and maintain trust. Used poorly, though, he warned that "you can damage the thing that is most valuable to you and your brand communities, which is the trust they have in the journalism that you create."

Gilbert also noted that it wasn't until very recently that there was proof about how effective generative AI could be.  "We sort of took it as an accepted truth," he said, "that you're more productive with generative AI, but we hadn't really measured it."

He cited a research study involving 59 law students from the University of Minnesota Law School who were charged with completing "four basic lawyering tasks, representing a range of common tasks for entry-level lawyers."  The assignments were graded by an instructor who did not know which solutions were produced with or without AI.  Gilbert said students who did not use AI scored 3.07; those who used AI scored 3.17.

Most interesting, though, Gilbert said the study showed that the use of AI didn't make as much of a difference among the highest-performing students as it did among those lower on the scale. 

"So, if I were running a news operation," he said, "that sort of thing — generative AI — may not be for my best reporters.  Generative AI may not be for my best editor.  But it may enable me to take people with less experience who are not as high-performing and raise the floor on their performance."

He stressed: AI won't replace your really talented people.  But, maybe it can help to make everyone else a little bit more like your stars.

He also discussed the work being done by the Knight Lab at Medill, which is studying what news needs to be in 2030.  For this study, they selected 45 young news consumers between the ages of 18 to 25 from the United States, Nigeria and India.  Regardless of what country they were from, whether they lived in urban or rural communities, or how much education they had, Gilbert said the researchers saw the same patterns.

They spent two to three hours with each person — in their homes, their place of work or a third place (like a coffee shop or restaurant) — wherever they felt most comfortable.  They asked the young news consumers to diagram their day, from the time they woke up to the time that they went to sleep.  They talked abut their information consumption (news, non-news, everything) and they asked them to detail their ideal news experience; not what it is now, but what they wanted news to be like. 

"Some people talked about a robot bringing them news," Gilbert said.  "Does anyone actually want a robot to carry the news to them?  Of course not.  But, it sort of spoke to the idea that they were not looking to have to go find things as they so often do today, but rather to have the news that they need brought to them."

Then, they were asked to open their phones, and show the researchers the apps on their home screen.  Gilbert acknowledged that this might be "the most intrusive thing we have ever asked of almost anyone." They asked the young news consumers to show them what they open, their screen time usage and how much time they spend on various apps.

Of the 45 people interviewed, one young news consumer told the researchers that she was paying for The New York Times, but despite having a college degree from a top American university, she said she doesn't feel "smart enough" to even understand the news from the paper she is subscribing to.  So, why subscribe?  She wanted to be able to open specific articles that people sent her links to — and she had been hitting the paywall.

Feeling flooded with the amount of information that is out there is something that the group as a whole cited.  The question became: How do you figure out what information is worth engaging with? Gilbert said the research showed that young news consumers are very savvy about algorithms, but probably think algorithms are more sophisticated than they actually are.  And, they described a huge sifting process to determine which notifications to pay attention to.

Gilbert said the study identified five modes of consumption among young news consumers:

  • Sift — How do I filter through the noise?
  • Substantiate — How do I know what's true?
  • Study — How do I build knowledge and expertise?
  • Socialise — How do I connect with others?
  • Sensemake — How do I figure out what it all means?

And, Gilbert noted that a large expectation gap exists between the supply of information coming from news creators and the demand side (what consumers actually want).  The gap between the two is really important, he said, because "it gives some opportunity for publishers to meet the needs of news consumers in ways that might make more loyal customers."

To meet this need, the researchers created a news framework with three components.  The first involves being a trusted source, which is an area where Gilbert said local news shines.  But, he encouraged publishers to be more transparent about their motives and how they finance the journalism that is produced. He said that will go a long way toward establishing trust, especially among young people who think news publications just want them to buy products shown in ads.

Second, credibility is tied to this, as well. "Can we tell people why we are the right messenger and report on the topics that we report on?  Are we deeply established in the community?" he asked. He encouraged newsrooms to seek ways to make the topics that communities need to know about interesting enough for them to want to read.  "And, then can we make it actionable?  Still being neutral, still being non-partisan, still being objective.  Can we give people a next step when they consume this information? What can they do with what we just told them?"

Finally, he encouraged newsrooms to be more flexible in their approach to storytelling and acknowledge that not everyone wants exactly the same version of the same story.  He said, "Not everyone who reads our journalism wants the kind of polished, professional version of the news.  A lot of people  want something that is more colloquial, more in their specific vernacular.  And, if we can find ways to meet them where they are in terms of the language that they want, we can build a better relationship with them."

Newspapers also have a lot of first-party data, he said, that they can leverage to better understand how long people will read at certain times of the day, and how they engage with emailed newsletters versus the website, for example.  This data can help newspapers find ways to make the information they gather available to readers in a variety of different formats.

Download the toolkit

To help newspapers implement the results of the research findings, Medill has produced a toolkit that can be downloaded.

The toolkit will help newspapers utilize generative AI's core technology in the right way, Gilbert said — as opposed to thinking about it as a search engine or trapping yourself into thinking generative AI is just a chatbot.

He stressed that generative AI will not be a replacement for human reporters.  Instead, it's a tool that can help reporters in a variety of ways, such as: with investigative reporting, in preparing questions for interviews, and taking the journalism that has already been done and making it available in additional formats (for example, restaurant guides). 

AI also can be valuable in identifying subconscious bias in news, he said.  As an example, he said it could be used to analyze a story and identify all the places that a reporter might have accidentally used biased language.  Ultimately, it would be up to a human reporter or editor to decide whether any changes suggested by AI should be made.

Historically, newspapers have been a one-to-many business, he said.  But research shows that the next generation of news consumers wants more.  "I hope generative AI will allow us to move from a one-to-many model to a one-to-one for many models," Gilbert said.

Download his PowerPoint presentation in PDF format

The Mega-Conference also featured additional sessions on AI:

  • AI is for More Than Content! Using AI to Enhance Operations — With Rachel See, senior counsel, Seyfarth Shaw
  • AI and the Future of Local Newspapers: How One Newspaper Group is Leveraging AI to Expand Hyperlocal Reach, Enhance Reader Engagement and Unlock New Advertising Opportunities — Moderated by Lloyd Armbrust, CEO, OwnLocal, with Matt Miller, CRO of Trib Total Media, and Joe Lawrence, general counsel of Trib Total Media and CEO of MeSearch