Journalism academics share recommended GenAI tools for newsrooms
Conference Blog | 06 June 2024
Newsrooms have plenty of tools to choose from when it comes to using generative AI, but how do you know which ones are right for yours?
During the Generative AI Master Class, part of INMA’s Generative AI Initiative, Jeremy Caplan, director of teaching and learning at the City University of New York’s Newmark Graduate School of Journalism, and Nikita Roy, a Knight Fellow at the International Centre for Journalists in Toronto, walked INMA members through a gallery of applications designed to make journalists’ jobs easier and more efficient.
YESEO is a free Slack-based tool designed for local newsrooms that’s now in use in more than 400 newsrooms. Users paste in an article and receive headline and SEO suggestions.
This is one of the tools that many people are most interested in, Roy said: “A lot of newsrooms have created them internally, but if you’re a newsroom that’s a smaller size and you don’t have the technical capabilities within your team to do it, this is a free alternative option for you.”
Google Pinpoint allows users to apply AI to a large collection of documents that has made it popular with many leading newsrooms, Caplan said. Users can extract text from handwritten documents and upload thousands of PDF files and search them.
“This could be audio files, or video files that you’ve transcribed, and you can have a huge volume of material that you’re then applying an analysis to,” he said.
Pinpoint will search for entities, organisations, people, and more. As proof of its effectiveness and efficiency, the Tampa Bay Times used Pinpoint as part of a Pulitzer Prize-winning project to analyse a massive number of documents: “It’s a really powerful, multipurpose kind of tool for newsrooms to add AI.”
Another Google tool useful for assembling large amounts of information is NotebookLM. The tool lets users upload notes and then use AI to synthesise and pull information, Caplan said.
“When I’ve collected a lot of notes on a particular topic I’m reporting on or researching … I use this to essentially synthesise all the kinds of information. Then I can have a dialogue with it and it pulls out patterns and connects material across your notes,” he explained. “It’s really a way to apply AI to a very custom set of material.”
Oasis is another tool that’s useful for keeping track of thoughts and ideas. Roy said she can record her ideas and choose how she wants AI to assemble those ideas — as a blog post, an e-mail, or even a song. She most often uses it for emails and brainstorming:
“I find it really helpful, especially if I’m doing emails or creating and drafting an outline of an article that I’m working on,” she said. “All I have to do is talk through my idea and Oasis is able to set up an outline for me that I can then go and build off of.”
For data journalism, Roy suggested Wobby AI, which allows users to ask questions about massive data sets and receive an easy-to-understand answer. This is particularly useful for newsrooms that don’t have a data journalist on staff.
“This was actually built by data journalists and it’s built for data journalism specifically,” she said. “It’s able to create all of these visuals and it automatically can be converted into a graph that you can import and embed within your site.”
When it comes to video and audio, AI offers multiple tools that can benefit newsrooms. Opus Clip can help convert long YouTube videos into short, catchy TikTok videos, making them ideal for social media promotion, Roy said.
“I find this really useful because a lot of us just put all of these Webinars and YouTube clips out on YouTube, and we need to be able to repurpose them for videos,” she said. “This is one way you’re able to get these quick shorts generated.”
One of Caplan’s preferred AI tools for podcasts is Snipd, which summarises podcasts using AI. It provides an overview of the episode’s content and allows users to share clips from that episode.
Plug-ins to existing programmes allow users to expand their capabilities with illustrations. Caplan discussed Design Mentor, a ChatGPT plug-in that works with Figma. Users upload Figma sketches and get feedback on the design from AI.
CanvaGPT works through Canva and will create whatever you tell it to, Roy said: “It will create it on Canva, then you just have to go and edit it, and it’ll create it and put on text and images.”
Drawn to Style is an image generation tool that allows users to customise the style of an image, while Sketchy creates black-and-white sketch drawings. This allows users to go beyond the traditional images generated with AI, Caplan said: “You can really get a lot more nuanced and have a consistent style for illustrations if you’re using AI-generated illustrations in your newsroom.”