From classrooms to the arts and every industry in between, generative AI has sparked countless conversations about how “the way we do things” is forever changed – and what humans need to do in order to adapt.
Joe Amditis, assistant director for Products and Events in the Center for Cooperative Media, is particularly interested in how newsrooms (particularly small operations with ever-tightening budgets) can best leverage AI. Here, he shares tools and tips for optimizing AI tools like ChatGPT, ethics concerns as it relates to AI and what he predicts for the future of local newsrooms and AI.
What sparked your interest in how AI could be used for local news?
I’ve always been interested in how local newsrooms, especially those with fewer than five full-time staff, can improve their operations and workflows to make them more efficient and effective, even on a shoestring budget.
I view generative AI as an opportunity for small newsrooms to continue that trend and find ways to automate some of the more tedious, monotonous tasks that newsrooms have to tackle each day and then use the time and energy they save to focus more on serving their community and producing quality journalism.
What are some ways local broadcast, digital or even print news publications could be using AI?
Generative AI can be used in a variety of ways by local newsrooms, including:
- Crawling and scraping the web for news stories. AI can be used to automatically find and collect news stories from a variety of sources, including websites, social media, and even live video feeds. This can help news organizations to stay up-to-date on breaking news and to identify stories that are relevant to their audience.
- Automating the production of news stories. AI can be used to automate the process of writing news stories, from gathering information to generating the text of the story. This can help news organizations to produce more content with less manpower.
- Personalizing news content for individual readers. AI can be used to personalize news content for individual readers based on their interests and preferences. This can help news organizations to keep readers engaged and to increase the likelihood that they will click on and read the stories that are presented to them.
- Targeting news stories to specific audiences. AI can be used to target news stories to specific audiences based on their demographics, location or other factors. This can help news organizations to reach a wider audience with their content.
- Measuring the impact of news stories. AI can be used to measure the impact of news stories, such as how many people read them, how long they spent reading them, and whether they shared them on social media. This information can help news organizations to understand what types of stories are resonating with their audience and to make changes to their content accordingly.
- Summarizing or repackaging existing content for different audiences and platforms. AI can be used to create shorter versions of articles for social media or other platforms with limited space; generate new headlines or summaries of articles to make them more engaging or relevant to different audiences; translate articles into different languages; create audio or video versions of articles; and/or generate new content based on existing articles, such as lists, quizzes or infographics.
What advice do you have for aspiring, early career or veteran journalists about why they might want to consider integrating AI into their work?
Start using it NOW. Even if you’re just playing around with the different tools, the time to familiarize yourself with the basic processes and tactics around these tools is yesterday. They’ve only been out (in a big way) for less than six months, and the landscape is constantly changing and evolving. The good news is, the interface is as easy as having a conversation. The bad news is, we have no idea what this space might look like in another six months, so start paying attention ASAP, or you and your company will likely be left behind.
What ethics issues or other concerns do you see emerging as more newsrooms embrace AI?
One of the things that I’m most worried about is the inevitable point when these technologies are adopted by right-wingers, fascists and bullies. These tools have the potential to make targeted harassment, scams and grifting much more efficient and much more personal – at a scale that was previously impossible. As things continue to change at such a rapid pace, we have to be prepared and ready to protect the most vulnerable among us.
Can these strategies be scaled up and used by bigger media companies? How?
These strategies can absolutely be scaled up and used by bigger media companies, especially if they are willing to dedicate resources toward projects such as fine-tuning models on their archives and community data, building custom chatbots to increase and improve engagement with members of the community, creating specific workflows and processes to streamline the production (and promotion) of existing content, and more.
- Web crawling and scraping. Use AI to collect news stories from various sources at a larger scale.
- Automated news production. Implement AI to streamline content creation, including writing, fact-checking and data analysis.
- Personalized content. Utilize AI algorithms to deliver tailored news to individual users based on their preferences and behavior.
- Targeted audience. Employ AI to segment and target news stories to specific demographics or user segments.
- Impact measurement. Analyze user data with AI to measure the impact of news stories and optimize content strategies.
- Content summarization and repackaging. Utilize AI for tasks like summarizing articles, translating content and adapting it for different platforms and audiences.
Which news publications are making good use of AI?
Scott Brodbeck at ARLnow is already finding interesting and useful ways to incorporate the use of generative AI to help create newsletters for the publication. You can read more about ARLnow’s efforts in this recent Nieman Lab article, in which I am also featured.
How has the Center for Cooperative Media experimented with AI?
We’re using AI to help summarize our content and activities (blog posts, news stories, events, etc.) and make them easier to promote and share. That includes suggesting alternate headlines, possible social posts and newsletter blurbs. We’ve also used it to help us generate emails, documents and other content that is personalized using existing data about the recipients or partners involved.
What do you predict for the future of local news as it relates to the impact of AI?
I think we’re going to see much less reliance on so-called “prompt engineering” in the future, especially as this technology is standardized and incorporated into various SaaS products and services. All of the major tech companies are already racing to join the AI space, and I think we’ll eventually see more of a shift away from prompt engineering, per se, and more toward “prompt management” or AI operations management.
As I wrote in my Nieman Lab prediction for 2023:
The ability to quickly generate documents, instructions, guides, and various elements of internal infrastructure could save independent and hyperlocal news publishers an incredible amount of time and effort.
Instantly generating summaries of public meetings and documents, creating tweets and social posts from news stories, drafting scripts for news broadcasts, even suggesting different headline variations – all at the click of a button – would be a game-changer for news organizations that are already strapped for people and resources. The same thing goes for generating invoices, public records requests, and even basic outreach emails.
That’s more time that could (and should) be spent reporting or engaging with community members. It could also help make local news content more accessible to people with disabilities, people who speak English as a second language, and people who might not usually come across local news stories in their current format.
I’ve also written about some of the larger, more big-picture possibilities of this technology once it matures to the point where fine-tuning models is as easy as uploading a set of files and documents to the web.