Tim O’Rourke is the vice president of Editorial Innovation and AI Strategy at Hearst Newspapers. With AI evolving at breakneck speed, the challenge for newsrooms isn’t using it, it’s integrating it responsibly so it enhances journalism rather than replaces it.

For many local and regional outlets, these tools bring both opportunities and challenges, making reporting more efficient while raising questions about accuracy, transparency, and ethics.

Below are excerpts from our conversation with O’Rourke, on how Hearst is approaching innovation, what makes newsroom AI trustworthy, and where he sees the biggest opportunities ahead.

How is Hearst deciding which AI experiments to scale across its newsrooms?

Our DevHub team serves as a central editorial innovation unit for all Hearst newspapers and markets. The group originated at the San Francisco Chronicle, where O’Rourke was managing editor, before becoming a system-wide resource to help local newsrooms adopt new tools and workflows. 

“Our mission is to help most journalists and serve the most readers,” he said. “Everything we do goes through the prism of how we can scale it across the group.”

When generative AI tools like ChatGPT emerged in late 2022, the team began experimenting quickly, adding them to a long list of projects in interactivity, data visualizations, automation and editorial engineering. Those efforts led to Producer-P, an AI assistant built inside Slack that drafts SEO headlines, checks readability and creates social-media variations while keeping editors in control.

O’Rourke said Hearst decided which projects to expand system-wide by weighing three key factors: whether a tool can be replicated across markets, whether it provides clear value to journalists or readers, and whether it offers something distinctive that pushes newsroom innovation forward. 

“Even if a prototype doesn’t become a home-run success, we learn from every experiment,” he said. “Each one moves us a little ahead in how we think about editorial innovation.” 

What design choices make AI tools most trusted and useful for journalists?

For Hearst Newspapers, trust and utility in AI design start with one rule — keep journalists in control. Basically outlying on “human-in-the-loop”approach.

“Our core principle is to keep humans in the loop,” O’Rourke said, “We’re not trying to replace editorial judgement. We’re trying to replace the rote and robotic work that comes with journalism.”

He described Hearst’s approach as “cautious innovation.” The company prioritizes accuracy and editorial oversight as it tests new ideas. “We’re in the accuracy business. Whether it’s an internal tool or an external experience, accuracy has to be guaranteed.”

When designing internal AI systems, Hearst focuses on usability and proximity to the journalist’s workflow. Producer-P, the company’s in-house audience optimization bot, was initially built in Slack, not directly in the content management system (CMS). “We wanted a bit of friction,” O’Rourke said. “AI tooling was so new that we wanted journalists to consciously review and paste content into the CMS themselves.” Now, as comfort with AI grows, Hearts plans to integrate such tools more closely with the publishing system while maintaining editorial review as a mandate.

For outward-facing products, O’Rourke said simplicity and utility drive design. “No matter the tool, the question is: Is it easy to use? Does it help the reader? We don’t want to make things too wonky. Simple is good.”

Asked whether veteran journalists fear AI might threaten jobs. O’Rourke said skepticism exists but isn’t generational. “It’s not about age, it’s about personality,” he said. “Some are early adopters, some are cautious, and some are resistant. Our job is to build trust and show that these tools can help without compromising standards.”

To illustrate, he pointed to Heart’s Assembly project, an AI-powered system that monitors public meetings, transcribes discussions and summarizes key points for reporters. “You can’t attend two meetings at once,” O’Rourke said. “This tool lets journalists skip the hours-long recordings and focus on the real world, reporting and storytelling.”

The goal, he added, isn’t automation for its own sake but freeing journalists to do what only humans can: verify facts, apply judgement and tell stories that matter.

What do you think is the biggest opportunity for innovation in the crossroads of AI and journalism in the next few years? And where do you see it?

“As quickly as this technology is evolving, it’s hard to look six months ahead, let alone years,” said O’Rourke. “There isn’t just one opportunity. This is a tool unique in its scope, and it will affect every facet of our industry.”

He sees early promise in investigative and data-driven reporting, where automation can take over tedious groundwork and free journalists for higher-level analysis. “If a reporter has to evaluate 100,000 gnarly-looking PDFs, a bot can do that first pass,” he said. “Then the reporter steps in to do the harder work, the interpretation and storytelling.”

Heart’s data teams already handle national datasets that are localized manually for each of the company’s markets. O’Rourke said AI could eventually streamline that process without compromising accuracy. “We do the cleaning, the analysis, the visualization, and then we can localize it, all by hand,” he said. “You can imagine a future where the machine handles some of that repetitive work, and the journalist ensures it meets our standards.”

Beyond reporting, O’Rourke said AI could simplify the hidden workflows that newsrooms created to adapt to digital publishing and social platforms. “If we can let our machines do the work to please their machines,” he said, “then we can stay focused on doing high-quality local journalism.” 

He also sees potential personalization, tailoring local news experiences to readers’ communities and interests. “We’ll never have enough staff to cover every single neighborhood in the Bay Area,” he said. “But can AI help surface the right set of stories for each reader? Can it understand that my needs in the suburb differ from someone downtown? That’s the direction we’re heading.”

How do you see AI changing the day-to-day work of reporters and editors? Do you imagine AI handling more transcription, headline drafting, data crunching while journalists have the free time to just interview and report? 

Artificial intelligence is already handling many of the repetitive tasks that used to fill journalists’ days, but O’Rourke says the goal isn’t to replace human judgement, it’s to make space for it.

“Right now, our AI tools — whether vendor-built or in-house — do a good job at the basics: summarization, SEO, tagging, keyword optimization, and building a briefing from vetted stories.You can still get better output from a skilled specialist, but the bots are getting better every day.” he said.

O’Rourke said AI’s immediate value lies in eliminating routine tasks that slow down reporters and editors. “If we can take a process that’s a pain for our journalists and get it out of their way, they can spend more time on the higher- level work, interviewing, verifying ,and storytelling.” he said.

But for journalism that reaches readers directly, he believes the human touch will remain essential. “On your highest-touch stuff, the stories, the headlines, the visuals that go all the way to the reader, you still want a human element,” O’Rourke said. “That’s vital when audiences are sifting through so much automatically generated content.”

One area he sees ripe for improvement is newsletter, a format that relies on curation and personalization. “Imagine if you could drop in your stories and the bot automatically sorted versions for different audiences, one for your city, one for New England, one for the East Coast, without you even having to think about it,” he said.

That, O’Rourke added, is the kind of innovation that excites him: technology that quietly supports journalists while helping newsrooms grow their reach.

What lessons did Hearst learn from the San Francisco Chronicle’s Chowbot that shaped the stronger San Antonio version?

When Hearst Newspapers rolled out AI-powered dining assistants at its San Francisco Chronicle and San Antonio Express-News, the tools shared the same foundation, but one clearly resonated more.

“In terms of raw traffic, they’re about equal,” he said. “But there’s no question the San Antonio version is a better product and a better user experience.”

The Express-News project builds on lessons from San Francisco’s earlier “Chowbot”, an AI navigated layer designed to guide readers through restaurant reviews and dining coverage. The key was centering the tool around the human journalist behind the work, in this case, longtime restaurant critic Mike Stutter. 

“Mike is unbelievably talented and respected in the community,” O’Rourke said. “We wanted his voice to be part of the experience. The tool exists to get people to his content more easily, the guides, the reviews, the news, without losing that human touch.”

In San Antonio, the redesign turned the chatbot into a true navigation engine rather than a novelty. “It’s about helping readers cut through dozens of guides and stories to find the best dining information in the region,” he said.

The revamp required building the product from the ground up, but O’Rourke said that effort will soon pay off Chronicle readers as well. “We did all the hard work with San Antonio — front end, back end, the whole thing, so relaunching the San Francisco Chronicle Version with those same bells and whistles will be a much smoother process.” he said.

Because the Chronicle relies on multiple food writers rather than one single critic, the next iteration will need a few tweaks. Still O’Rourke believes both versions share the same goal: helping readers connect more intuitively with the journalism that makes local journalism shine.

What is your advice to students just out of school who are interested in this intersection of journalism and tech, and just using AI within the newsroom to enhance traditional journalism?

For students and early-career journalists eager to work at the intersection of technology and storytelling, Tim O’Rourke’s advice is simple: learn actively and seek out mentors.

“It’s never been easier to learn on your own,” he said. “From online resources to coding assistants and AI tools, you can fill in your skill gaps and really own the process.”

But, he added, self-learning should be paired with real newsroom experience. “At the same time, it’s about getting into places where you can learn from people who are really really good,” he said. “Find internships, fellowships, and organizations that are doing it right, the ones that respect journalistic ethics but also prioritize innovation.”

O’Rourke said that combination, curiosity, technical fluency, and exposure to strong newsroom culture, will help new journalists thrive as the industry evolves. “There are a lot of smart people in this space now,” he said, “It’s just about finding your entry point and building from there.”