As AI replaces apps and the web to become the main interface for accessing online information, journalists can no longer wait to learn about, experiment or simply just play with AI. Journalists cannot afford to let the technology define journalism. Instead, newsrooms must actively participate in shaping the role of AI in journalism, defining what AI means to them and setting clear benchmarks of excellence.

These themes were repeated over two days of discussions and talks during Austin’s annual SXSW megaconference, at AI x Journalism House, held March 10-11 at The Texas Tribune. The event, produced by Hacks/Hackers in partnership with the Online News Association (ONA) and with support from MacArthur Foundation, brought together industry leaders for workshops, discussions and networking at The Texas Tribune. More than 100 attendees, including journalists, news product managers, researchers and others came to listen to and learn from presenters including The New York Times, Der Spiegel, Bluesky, The Associated Press, and ONA about how AI is reshaping journalism.

In an era of AI, audiences still value the role of journalists

The public thinks of journalists as an "iconoclastic person who loves to ask the hard questions," something that "AI is not good at,” said Celeste LeCompte. This means that even as AI becomes more and more ubiquitous in journalistic workflows and as an interface, audiences still value the skepticism and critical inquiry of actual human journalists.

LeCompte, a contributing expert with the Center for News, Technology & Innovation (CNTI), a policy research center focusing on independent, sustainable media, was presenting together with CNTI executive director Amy Mitchell about the results of two recent surveys that explore how journalists and the public around the world perceive journalism and AI. 

A striking insight from the CNTI surveys is that, at a time when news influencers continue to dominate the media ecosystem, at least half of both journalists and the public say people who are not journalists can produce journalism. Mitchell and LeCompte’s research also shows that, while two-in-three journalists think that digital technology will have a positive impact on their ability to report, only one-in-three think AI will have a positive impact on the public’s ability to be informed.

During their talk Mitchell and LeCompte discussed how it actually isn't about technical capabilities of AI. The question is “are we meeting audience needs through the use of these technologies or not?” said LeCompte. 

According to the focus group participants LeCompte spoke with as part of the two CNTI surveys, audiences devote significant effort to keep track of what's happening in the world. “They want to reduce the amount of work they have to do to get information fast. They want it to be clear and easy to read. And they really want it without what they perceive as bias.”

Generative AI tools such as Gemini and Microsoft’s Copilot “tick all those boxes” for what people want from news, said LeCompte. “It's not that AI is doing a better job of answering the question correctly. It's that it's meeting their need for how they want the information very fast.”

Meanwhile, there's a noted historical resistance and caution among journalists in the US towards adopting new technologies, potentially putting journalism at a disadvantage, said Mitchell, even as there is a critical need for journalists to embrace and become knowledgeable about AI.

“Journalists have to be able to take some time to just play and learn and become experts in both what they can do but also what others are doing and what's happening”, said Mitchell.

Audiences don’t care about the tech behind “the refrigerator,” they want “the vegetables” inside it

Stefan Ottlitz, Co-CEO of Der Spiegel used the analogy of a refrigerator to describe the German weekly news magazine’s approach to deploying AI.

“AI will become as commonplace and unremarkable as a refrigerator in the kitchen,” said Ottlitz. “People will accept its presence without much thought about the underlying technology. The ‘vegetables inside the fridge’ are what truly matters.” Or, in the case of Der Spiegel, high-quality journalism.

Christoph Zimmer Der Spiegel’s chief product officer, stressed artificial intelligence is not replacing journalists, but instead enhancing their efforts.

“We're not focusing on using AI for content production,” said Ottlitz. “It's not about putting out more content out there but instead to make it easier for journalists to produce (journalism).” He emphasized that the magazine’s journalism and its brand are more important for engaging with and growing its audience; AI should serve journalism rather than redefine it.

Zimmer said his goal is to integrate AI across operations, including using AI to fact-check against Der Spiegel’s vast archives, using AI for training, developing newsroom tools and experimenting with chat-based interfaces that allow users to interact with decades of content. 

Artificial intelligence is also key for navigating an era of “liquid content,” where journalistic content exists “in bits and pieces all over the Internet,” including as search results in AI chat interfaces, or in augmented reality interfaces such as audio and in earbuds. There’s a need for Der Spiegel and other publishers to proactively preserve their content’s integrity wherever it flows, Ottlitz and Zimmer said, rather than passively allowing it to happen without their influence.

How to successfully lead AI adoption in the newsroom

Resistance to AI adoption in newsrooms is a common experience, said Aimee Rinehart, Senior Product Manager AI Strategy, AP. "It's all resistance – nobody wants to do something different." Addressing this skepticism to AI adoption requires that news product teams demonstrate the potential benefits of AI to journalists.

Rinehart — together with Rubina Fillion, associate editorial director of A.I. Initiatives at The New York Times and Elin Wieslander, AI and News Personalization Editor for  Sweden’s Aftonbladet — discussed how news product managers actually lead and facilitate AI adoption in newsrooms.

Starting with focused experiments and developing a proof of concept is a key strategy for implementing AI in newsrooms, according to the panelists. Rinehart suggested starting with "tiny experiments" with small teams over limited timeframes to deliver results and demonstrate the value to leadership. Fillion echoed this, describing how she runs small pilots with different desks across The New York Times to gather feedback and show proof of concept before wider implementation.

Rather than replacing people in the newsroom, AI should be used to help journalists save time, said Fillion. “Our SEO team would sometimes have to write different headlines for 10 different pieces published on an election night that all use the same keywords… which is really hard,” Fillion said. An AI-generated “cheat sheet” for quickly extracting information from large lists of keywords makes brainstorming much easier. 

Building AI news product at The Texas Tribune

Darla Cameron, chief product officer of The Texas Tribune, shared that her team has sometimes struggled to build AI tools that work as intended, or are embraced by journalists and audiences alike. For example, when the Tribune decided to train a chatbot using 15 years of the paper’s archives, Cameron and her team encountered several issues.

"Despite our best efforts to prompt the tool, it would sometimes like totally forget about our archives, what it was supposed to be doing,” said Cameron. “A bigger problem was that the chatbot didn't know about recency, and ‘new’ is part of news, right?” 

Cameron used the example of Texas attorney general Ken Paxton who was impeached in 2023, and was briefly out of office before being reinstated. When asked, however, the Tribune’s experimental chatbot could not determine if Paxton was in fact in office. The issue, Cameron said, was that the model relied on the volume of archived stories produced during the leadup to the impeachment trial, and not the conclusion of the story – Paxton's return to office as attorney general.

“The tool was also pretty expensive to run, even internally,” said Cameron. “Our journalists complained about needing to go to a specific website to use it, and it had pretty low uptake among staff." In response, Cameron's team pivoted to build in-house bots in Slack, an environment more familiar to the organization.

Cameron said the Tribune either buys or builds its AI tools, or partners with others to get what it needs. For example, the Tribune provides its reporters with access to a premium version of Perplexity AI, which can be more effective than other search tools for locating government files. Cameron’s team has also built an in-house Slack help bot that has been trained on the company handbook and other internal documentation to quickly answer staff questions and improve efficiency.

Cameron also shared the story of a Tribune reporter who used the tool Gigafact to analyze over 500 episodes of Ted Cruz's podcast, which, she said, would have been impossible to do manually.

“Ted Cruz has a podcast that comes out three times a week since 2020,” Cameron said. “If (the reporter) had sat down to listen to each podcast, he would not have had time to do any other work. This story came out during the election cycle, so (using an AI tool) helped the reporter to quickly get a sense of the tone, the breadth of what Cruz talks about on his podcast.”

Human journalists are still important in the AI era

A unifying theme throughout the various talks, panels and presentations at AI x Journalism House 2025 was that the goal is to empower journalists by using aI to assist with often mundane daily tasks, helping their ability to investigate and report by using powerful new tools.

As journalistic content becomes a “liquid” that flows in bits and pieces all across the Internet, including AI chatbots, summarization apps and voice assistants, newsrooms need to experiment with transforming their content into various formats such as audio, video and interactive elements. As content becomes increasingly fragmented and distributed across various platforms establishing direct relationships with users, establishing trust based on the core journalistic values of accuracy, in-depth reporting, and ethical standards.

Join Hacks/Hackers in Baltimore for the AI x Journalism Summit

​Join pioneering journalists, technologists and innovators May 7-8 for two transformative days exploring how AI can enhance journalism and information. Through hands-on workshops, real-world case studies, and collaborative sessions, discover how to leverage AI to strengthen reporting, streamline workflows, create more impactful stories and build innovative new products.

So far over 30 speakers will present on three tracks:

• ​Reporting on AI: With AI being integrated into more parts of everyone’s lives, journalists will play an essential role in documenting the effects of algorithms on people’s lives and holding institutions accountable
• ​Using AI: Learn practical ways to incorporate leading AI tools into your workflow - from automated transcription and translation to using large language models for research and analysis. 
• Building with AI: What do future media products look like when they leverage the power of generative AI? How will we reimagine user interfaces, discovery, distribution and monetization of news and information products?

Sessions and Speakers from AI x Journalism House 2025

If, When & How to Use AI in Journalism – and Tell

• Amy Mitchell, Executive Director, CNTI
• Celeste LeCompte, Contributing Expert, CNTI

AI, Innovation, and the Reinvention of Legacy Industries

• Amanda Edmonds, VP of Business Partnerships, ProRata.ai
• Virginia Fletcher, CTO, Lee Enterprises

Trust & Safety in Generative AI

• Lavanya Poreddy, Head of Trust and Safety, HeyGen

Real Talk About AI's Newsroom Revolution

• Aimee Rinehart, Senior Product Manager AI Strategy, AP
• Rubina Fillion, Associate Editorial Director of AI Initiatives, The New York Times
• Elin Wieslander, AI and News Personalization Editor, Aftonbladet

Sentient Design: Shaping Literacy Through Design in the AI Information Age

• Veronika Kindred, Associate Producer, Big Medium

Featured Conversation: Independent Journalism Despite, With, or Thanks to AI? The Der Spiegel Perspective

• Stefan Ottlitz, Co-CEO, Der Spiegel
• Christoph Zimmer, Chief Product Officer, Der Spiegel

Lessons from The Texas Tribune: Building AI Tools for Accessibility

• Darla Cameron, Chief Product Officer, Texas Tribune

Interrogating Generative AI Through Community Juries

• Meme Style, Founder & President, Measure

Featured Conversation: How Bluesky is Reimagining Breaking News and Journalism

• Rose Wang, COO, Bluesky

Reimagining Roles for People in the AI-Powered Newsroom

• Hanaa Rifaey, Interim Executive Director, Online News Association
• Meghan Murphy, Director of Programs, Online News Association

From Archives to Audiences: How AI is Powering the Next Era of Content Discovery

• Kevin Hill, General Manager, Media & Entertainment, Coactive AI

Preserving Journalism in the AI Era: Trust, Provenance, and Decentralized Storage

• Rachel Horn, Chief Marketing Officer, Filecoin Foundation
• Brad Keoun, Founder and CEO, Distro Media
• Austin Carson, Founder & CEO, SeedAI