AI is rapidly transforming newsrooms everywhere, offering tools for journalists that can assist with research, images, planning tasks, and more.
Organizations like the Associated Press, The Haitian Times, and The New York Times are sharing how they use AI as well as ethical considerations.
Paul Cheung, an AI strategist and leading media executive, shared how journalists can integrate AI into their daily work during a National Press Club Journalism Institute workshop in May.
“AI is just a tool. You need to pick the tool that fits what you are doing,” Cheung said. “So, based on the job you are doing, you have to ask yourself which of these AI tools might make the most sense for you.”
Here’s how journalists can integrate AI into their daily work more effectively.
Customize your AI system: Under your ChatGPT profile, find the “Customize ChatGPT” tab. This feature allows you to personalize your AI experience by entering your profession, adjusting the tone of responses, and sharing additional information about yourself. Cheung encourages customization of ChatGPT because it “creates an expectation you have for the chat.”
Make AI fact-check its work: When using AI for research development purposes, breaking down the step-by-step process leads to more thorough results. The “chain-of-thought” prompting technique encourages you to ask AI to think before producing an answer, provide reasoning, and avoid any fabrications of data. After reviewing the output, ask the AI bot to rate its confidence on a scale of 1-10 and to identify any weaknesses or gaps.
Ask AI to challenge your perspective: It is important to create coverage from a nuanced perspective. Prompting your AI tools to provide counterarguments, review a story from different characters in your reporting, and highlight missed viewpoints can identify fault lines. Use AI to assist in pinpointing bias by asking how a specific group would frame the same story.
Let AI handle the document work: Claude, ChatGPT, and Google’s Notebook LM allow you to upload multiple documents and assist with summarization and highlighting main bullet points. As a pro tip, split your documents into sections and analyze one snippet at a time so the data isn’t so daunting and won’t take as long to process. With multiple documents, Notebook LM allows you to stay on top of connections over time. ChatGPT and Claude are better for a side-by-side comparison. (See how Bloomberg used AI-driven data analysis for an investigative journalism project.)
AI can help you become a wealth of knowledge: Instead of Google, Cheung uses Perplexity because it creates a contextualized search. When using Perplexity for research, users are granted an AI summary along with source links.
While AI is helpful, Cheung advises editing the work AI produces. “Never go from input, output, to publish. It’s always input, output, edit, augment, and maybe publish,” he said. “As these tools progress, it is important to fact-check to remain accurate.
In addition, check in regularly with your organization on updates to AI policies.
Interested in learning more about AI? Contact the Institute to gain access to a recording of our AI for Journalists workshop.