WELCOME

Get your newsroom ready to incorporate technologies that include artificial intelligence to support and grow all aspects of your news operation.


AP walks you through how to build a technology strategy across newsgathering, production, distribution and business operations.


The course is based on findings from AP’s research with local U.S. newsrooms and is designed for local news journalists and managers at all levels.


This course includes:

  • Guest speaker sessions
  • How-to videos of technology demonstrations and news applications


The learning experience was offered as live sessions in April and May 2022. Some videos are not included here due to rights restrictions.


This program is made available by funding from the John S. and James L. Knight Foundation.


Play Button

AI Readiness: Helping local news journalists understand AI

Live Session

Play Button

The AP Local AI team presents an overview of the role of artificial intelligence in journalism.

  • Jim Kennedy, AP’s senior vice president for strategy and enterprise development, describes how the organization incorporated natural language generation, or NLG, into its daily production process. Kennedy describes how narrative text transformed AP’s business news output by providing quarterly earnings reports for every publicly held company in the U.S. stock market.
  • Meredith Broussard, associate professor at NYU, explains five things that local journalists should understand when it comes to AI. Broussard also builds on Kennedy’s remarks, saying NLG is a computer filling out story templates and that journalists should not be intimidated by AI. Professor Broussard adds that many local journalists are already using AI-powered tools such as automated transcription.
  • AP’s Aimee Rinehart and Ernest Kung, who conducted a national research project on local news and AI readiness, present recommendations for local newsrooms looking to adopt automation and AI.


Convincing decision makers to invest in AI

Play Button

Jeremy Gilbert, professor at Northwestern University, explains what it takes to convince newsroom decision makers to pursue projects that involve AI. Gilbert, who previously worked at The Washington Post, observes that some decision makers will have a hard time understanding the value of AI. Journalists should plan a narrative story that creates a compelling case for the decision maker to pursue AI. Gilbert says pitches need to be focused on well-defined needs, whether those needs originate as stated needs or implied needs.


Crafting a pitch for AI in your newsroom

Play Button

Cynthia DuBose, managing editor for audience engagement at McClatchy, describes how to craft pitches to a leader. DuBose recommends starting with identifying gaps in your processes, and how automation can streamline or deliver additional content. Issues that must be addressed in a pitch include: potential time savings, project timing, audience impact and costs.


Taking the first step with simple process automations

Play Button

Adriana Lacy, digital and audience engagement editor at the Nieman Foundation for Journalism at Harvard University, demonstrates how to use IFTTT and Buffer to automate posting to various social media channels. Lacy, previously with the Los Angeles Times, walks through a process where new stories published by a newsroom via a RSS feed will automatically be sent to Twitter.


Newsgathering: Making it easier to gather facts

Live Session

Play Button

A trio of experts provide an overview of automation and AI-based tools that help in the newsgathering process across all formats.

  • Hilke Schellmann, assistant professor at NYU, provides a look at the Gumshoe application which is being integrated into DocumentCloud. Gumshoe processes large volumes of emails, such as those from FOIA requests, and determines if they might be relevant to your reporting.
  • Michelle Minkoff, an AP data journalist, reveals some insights discovered from AP’s reporting work across the U.S. Minkoff also covers how to identify tools that meet newsgathering needs and explores uses for natural language generation, or NLG.
  • Sophie Rosenbaum, news editor and audience strategist for AP, describes using automated tools to make coverage decisions. Rosenbaum also shows how Google Trends can be used to optimize story headlines to attract more viewers.


Getting a head start with transcribing interviews

Play Button

Elise Kaplan, a reporter for the Albuquerque Journal, demonstrates how to use the AI-powered Otter to automate transcription of interviews, news conferences and other recordings. Kaplan demonstrates the features of Otter, including ingesting raw content for transcription, starting the transcription process, editing the automated transcript, and sharing the transcript with colleagues.

Generating automated alerts from government websites (simple)

Play Button
Play Button

Samantha Sunne, a freelance journalist, demonstrates how to use Google Alerts and Wachete to generate alerts from government websites. Sunne walks through how to configure Google Alerts for very specific changes, such as new PDFs, Word docs, Excel spreadsheets, when they’re posted by a government agency. Critically, Google Alerts only updates once a day, so it’s not useful for breaking news. That’s where Wachete comes in. A paid tier offers checks hourly or sooner. Wachete can detect changes within a document in situations where governments release data using the same file name repeatedly. Wachete offers more options for journalists comfortable working with HTML.

Generating automated alerts from government websites (advanced)

Play Button
Play Button

Samantha Sunne, a freelance journalist, demonstrates how to use DocumentCloud’s new add-ons feature to generate alerts from government websites. This method allows for a combined process that includes automated monitoring of a government website, detecting new updates, importing new documents into DocumentCloud, and sending alerts based on monitored keywords. This method benefits from DocumentCloud’s abilities to run optical character recognition (OCR) on documents.


Avoid information overload on social media with automated tools

Play Button
Play Button

Samantha Sunne, a freelance journalist, demonstrates how to use TweetDeck and NewsWhip Spike. On TweetDeck, Sunne shows how to use filters in columns to focus on specific subjects, dates and locations. In NewsWhip’s Spike, Sunne walks you through searching for posts related to a specific breaking news event. Alerts can be generated on both products.

Production: Putting the facts together into a news product

Live Session

Play Button

These three experts provide an overview of automation and AI-based tools that help in the production process across all formats.

  • Anita Li, founder of The Green Line and a journalism instructor at Toronto Metropolitan University, discusses when it’s a good idea to automate newsletters when trying to reach an audience that is new to your brand. Li also covers using templates to speed up social media content creation using Canva.
  • Justin Myers, AP’s data editor, looks at automated text generation or natural language generation (NLG) and says that the AI behind these tools works in a specific way by inserting data into text. Myers introduces the Gannett team who explain implementation of automation to create thousands of data-driven COVID-19 stories.
  • Heather Edwards, AP’s director of metadata and data science, describes why it’s important to tag content that is related to each other. Taxonomy is a controlled vocabulary that sets up the relationships between different topics, but Edwards says that for newsrooms that publish low volumes, it is not cost effective to use AI to handle taxonomy. However, natural language processing tools can be valuable for high-output newsrooms and also to tag archives.


Play Button

AI-generated stories can free up journalists’ time

Play Button

Rich Scinto, data editor for Patch.com, demonstrates the features of Wordsmith by Automated Insights which automatically writes stories. Scinto explains that AI-powered writing requires a structured data set, which commonly appears as a spreadsheet. Patch uses Wordsmith to generate thousands of stories on COVID-19 data that are localized by county. Scinto shows in detail how he programs Wordsmith to fill in a story template using the data set. Scinto says Wordsmith has freed up time for journalists who would otherwise have to write repetitive COVID-19 data stories.

Distribution: Getting your news products to audiences

Live Session

Play Button

Three experts provide an overview of automation and AI-based tools that help in the distribution process across all formats.

  • Tim Kelley, director of digital publishing at Lee Enterprises, provides an overview of AI in comment moderation. AI-based tools can automatically review and approve comments posted about news stories, reject toxic comments, while leaving remaining comments for human moderators to review. Lee Enterprises uses a solution from OpenWeb and Kelley shows us the features of the AI-based system.
  • Carrie Brown, associate professor at CUNY, examines the wider view of engagement, distinguishing between audience engagement and community engagement. Brown says there are pitfalls to just relying on metrics for an audience engagement strategy. Community engagement can be time consuming but offers powerful potential over the long term.
  • Derrick Ho, deputy editor at The New York Times, shows how the Times is approaching the complex task of personalizing content for its audiences. Ho’s work focuses on the home screen using algorithmic programming to customize what certain readers see based on signals from them.


Play Button

What Google Analytics 4 can do for your newsroom

Play Button

Brad Gerick, a digital strategist at the American Press Institute, demonstrates the main features of Google Analytics 4, which all users will be required to transition to by July 2023. Gerick looks at the key features for newsrooms that have never used Google Analytics, and the features that current users will see in the new version. Gerick warns that Google Analytics should not be used for real-time decision making.

Using Chartbeat for real-time decision making

Play Button

Julie Gill Moreno, executive producer for digital at KSAT-TV, demonstrates the main features of Chartbeat. Moreno notes that the station relies on Chartbeat for the pulse of the day while using Google Analytics for deep dives on historical data. Editors can use Chartbeat’s visualizations to make decisions about how to showcase stories on the website.

To automate or not to automate comment moderation?

Play Button

Andrew Losowsky, head of community product at Vox Media, explains emerging best practices for audience comments on websites, along with providing an overview of the AI-powered Coral tool to moderate comments. Losowsky says any newsroom that intends to completely outsource comment moderation to AI should simply not allow comments at all. Coral pre-filters toxic comments, and offers feedback to commenters if they violate rules.

Business: Getting paid for your journalism

Live Session

Play Button

Three experts provide an overview of automation and AI-based tools that help in business operations across all formats.

  • Mike Orren, chief product officer at The Dallas Morning News, presents a case for using business intelligence in both news and business. Orren goes into detail on how the organization has segmented their coverage area. Certain cities get more reporters because they have the most potential for new subscribers. The News is looking to use AI to help cover areas that have less potential for new subscribers.
  • Mariah Craddick, product lead at The Atlantic, provides an overview of using a dynamic paywall to increase revenue. Dynamic paywalls, also known as adaptive or flexible paywalls, present a sign-up prompt if the AI predicts a user might subscribe. In other scenarios, the paywall lets them access the content for free. Newsrooms can decide to build a dynamic paywall in-house or seek a third-party vendor.
  • Max Willens, senior analyst at Insider Intelligence / eMarketer, covers the advertising technology market. Willens estimates publishers are getting one sixth of the money spent on programmatic advertising. The concepts can seem obscure, so Willens provides an overview to help newsrooms understand how the ad tech ecosystem works.
Play Button

Chatbots for customer and member service

Play Button

An overview of best practices for chatbots, along with a behind-the-scenes look at AP’s own customer service chatbot. Joe Germuska, executive director of Northwestern University’s Knight Lab, says chatbots can be built with no-code conversation flow diagramming tools all the way up to full-service consultancies. Tracey Rogers, manager for business and customer operations projects at AP, describes why AP settled on the Salesforce Einstein chatbot. Austin Cantrell, senior Salesforce administrator at AP, presents the user interface of the customer service chatbot, along with the flowchart-based programming interface. Cantrell says most of the chatbot was built without any coding.

Future State: What’s next in AI for journalism

Live Session

Play Button

The AP Local AI team conclude the six-week series with some predictions for the future of artificial intelligence in journalism.

  • Jim Kennedy, AP’s senior vice president for strategy and enterprise development, sits down with Francesco Marconi, founder and CEO of Applied XL, to discuss the evolution of AI in the newsroom. They describe using the news value chain of newsgathering, production and distribution, to guide prior AI initiatives at AP. The discussion looks at the role of computational journalists who can help filter and understand the massive volumes of data being generated by society. That work will require AI to filter out irrelevant data.
  • Nick Diakopoulos, associate professor in communication studies and computer science at Northwestern University, says the future of AI in news requires news values to be baked into algorithms. Journalists and news organizations have to understand their own values to apply them to AI. Diakopoulos says AI is rarely going to substitute a trained journalist, but in fact, AI will create new beats for journalists to cover and investigate. Diakopoulos says journalists need to be savvy enough to hold the AI algorithms accountable.
Play Button

Reporting on AI

Play Button

A panel of reporters who’ve covered the impact of AI in society describe the coverage opportunities in this space. Garance Burke, a global investigative journalist at AP, describes the real world impact of AI across multiple beats. Burke reported on a series called “Tracked” which examined how an AI-based technology landed a man in jail. Kathleen McGrory and Neil Bedi, reporters at ProPublica, describe their Pulitzer Prize winning reporting on AI in policing. Their series, “Targeted” for the Tampa Bay Times, examined how a local law enforcement agency used AI-generated information to harass certain people in the name of preventing crime. The three reporters explain how local newsrooms can begin to cover AI in their beats, along with recommending resources to help.