“AI changes everything,” is the conventional wisdom in and beyond local journalism, and it’s likely true.

The problem is we can’t change everything at once. As a result, many news organizations find themselves overwhelmed by the implications — both possibilities and threats — of artificial intelligence applications. So they do nothing. That’s not a competitive response to disruption.

What if you could develop an AI strategy for your organization in 30 minutes?

At the annual conference of the Colorado Press Association, I led a mix of 30 news leaders in conversations that set out to do just that. This group included managers, reporters, technologists, academics and student journalists, and they tackled the challenge of coming up with a strategic plan for newsroom use of AI.

In three teams of 10, this group followed a structured ideation process that produced 30 ideas for how to use AI to improve newsroom efficiency, enhance journalism and serve their community better. They also generated guidance and guardrails for when and how to use AI, and they identified the core practices they’d use to be transparent internally and externally about those uses. In 30 minutes!

Their insights are summarized below. The tactics are specific and actionable; more importantly, the process is repeatable in any news organization. The lesson? Asking the right questions, having a participatory process and including diverse voices can generate immediate, thoughtful and impactful actions that an entire team supports.

Here’s the four-step process we followed, and the takeaways that emerged.

1. Get diverse stakeholders in the room

Each of our groups included a range of perspectives, from early career to veteran journalists, news leaders, as well as those adjacent to journalism like technology partners, academics and community leaders. Each group included 10 people, which is a good size for a discussion. It’s large enough to ensure key ideas don’t get missed but small enough to ensure everyone gets heard.

2. Ask focusing questions

For a field as massive as artificial intelligence, it’s easy to get overwhelmed or to go down rabbit holes. To make the absolute best use of each person’s time, we asked only three questions — about use cases, ethics and transparency — and the questions were highly focused:

  • What’s ONE great use of AI to better serve audiences and sustain news?
  • If YOU were in charge, where is the line: What’s OK for AI and what’s not?
  • In a sentence, what do you tell audiences and the news team about your use of AI?

3. Everyone gets heard

To ensure the maximum number of ideas and perspectives were shared, and everyone got heard in a time-constrained setting, we used a practice I learned through IDEO at Stanford’s design school: Everyone received a small index card (no room to write an essay!), and for each of the three rounds of five-minute discussion questions, individuals wrote down their responses to the question prompts on the card. For each question, we went around the table so all 10 participants were heard.

4. Discuss, then narrow

Too often in group discussions, a few voices dominate while other contributors with a great idea might not speak up. Or there’s a rush to consensus. Using the process above, all participants were heard before we narrowed to a consensus on hard but important questions like: “Where do you draw the line on ethical use of AI?” and “What do you tell your audience — and when — about your uses?”

Best immediate opportunities for AI uses in news

(Editor’s note: ChatGPT was used to summarize the key themes, from 30 ideas collected by participants because one takeaway was: “summaries” are a valuable use case.)

  • Automating routine newsroom tasks: Tasks like spell-checking, grammar improvement, SEO optimization, and headline suggestions can be streamlined with AI, freeing up time for more in-depth reporting.
  • Interview transcription and analysis: Automating the transcription of interviews, saving time and improving accessibility for journalists.
  • Transcription and summarization of public meetings: AI can record, transcribe, and summarize public meetings such as city council or commission sessions to provide easily digestible key takeaways.
  • Translation and accessibility: AI can translate content to serve non-native speakers and improve accessibility, such as evaluating sites for compliance with accessibility standards.Data Analysis and Mining: AI can help analyze large datasets, spot trends, and provide insights, enabling data-driven journalism.
  • Content versioning: AI can repurpose content from articles into videos, audio versions, or social media posts, increasing reach and engagement.
  • Research and archival retrieval: AI can assist in deep research, including contextualizing stories with archived content or providing background information quickly.
  • Fact-checking and bias detection: AI tools can assist in fact-checking articles and detecting potential bias, ensuring accurate and balanced reporting.
  • Audience feedback organization and moderation: AI can organize, moderate, and analyze feedback from comment sections or community input, helping to understand audience sentiment and needs.
  • Legal notices and regulatory monitoring: AI tools can comb through legal notices or regulatory databases, surfacing relevant stories or tips for journalists, such as those found in the Federal Register.

Guardrails and guidelines: What’s OK and what’s not

Across the three discussion groups, a few key themes emerged for where and why to draw the line on uses of AI in local newsroom, including:

  • A human must always have the last touch on content
  • Journalists can’t assume the AI got it right — must fact-check
  • If you didn’t write it, you need to disclose it
  • AI use must be disclosed when it’s used for “meaning-making”

This part of the discussion was informed by the consensus among participants that many people in newsrooms were already quietly and unofficially using AI personally. So a proactive discussion and policy were essential to avoid misuse of AI.

Transparency: What we say (internally and externally) about our use of AI

Among the groups, several participants point to research showing that newsrooms that tell their audiences they are using AI are “punished” for their transparency. This led to a discussion about journalists’ role in promoting AI literacy among their audiences. Despite these concerns, participants agreed that transparency around AI use was a core journalism value and a few guidelines emerged:

  • If you didn’t write it, you need to disclose it
  • There’s a “duty to disclose” whenever NOT telling the audience would harm trust
  • Newsrooms should publish/post their AI policy for readers/users to see.

The key takeaway around transparency was that journalists and local news organizations must be responsible gatekeepers to ensure trust in our journalism.

A technology as transformational as artificial intelligence requires both immediate and ongoing engagement from all levels in a news organization. The methods described here can be used by any news outlet to initiate and maintain the ongoing dialogue necessary to leverage the opportunities and manage the risks of AI uses in news.

Read more about Ethical AI for newsrooms