SDS 910: AI is Disrupting Journalism: The Good, The Bad and The Opportunity

Podcast Guest: Jon Krohn

August 1, 2025

In this Five-Minute Friday, Jon Krohn looks into AI’s disruption of the journalism industry and how it has fundamentally reshaped news production. Multiple news outlets’ suing of ChatGPT over its use of copyrighted materials may have taken the most headlines to date, but this isn’t to say news media is rebuffing AI entirely. On the contrary, several outlets have launched summarization and analysis tools for both internal and external use, such as The New York Times’s Echo and The Washington Post’s Haystacker. This episode looks into the ways major news outlets are utilising AI, and what this means for journalists.


Summary writing and analysis appear to be the biggest use of AI in news outlets to date, and Bloomberg’s AI research team published four papers at the Empirical Methods in NLP Conference in 2024. Some editors also report significant time savings in transcription. But some organisations are also using AI to generate front-page stories, Norway’s iTromsø and Germany’s Express being two notable examples. 

Over eighty percent of over 200 respondents to a survey conducted by Thomson Reuters reported using AI tools in their work, ranging from drafting assistance to fact-checking and research support. Nevertheless, it appears AI in journalism has its limits, especially concerning text generation. The BBC found “51% of all AI answers to questions about the news were judged to have significant issues of some form” (BBC 2025), CNET’s AI-generated article on compound interest was riddled with embarrassing errors, and reports from academia reveal a reduction in nuance within AI-generated news articles. These issues might suggest that public trust in AI-generated information in the news is unlikely to improve.

And yet, the Reuters Institute recently noted some geographic differences in the use of generative AI as a news source. They found countries that were less skeptical of AI using chatbots to generate news adopting the tools more readily: “Almost a fifth (18%) of our Indian sample said they were using chatbots such as ChatGPT and Google Gemini to access news weekly, with comfort levels of 44%.” With the European Union’s AI Act and California’s AI Transparency Act (soon to be) in effect, these geographic differences in trust and use of AI may only widen in the years to come.

Find out exactly how AI is being incorporated into journalistic activities, and how Jon thinks journalists can future-proof themselves, in this week’s Five-Minute Friday.

ITEMS MENTIONED IN THIS EPISODE 


DID YOU ENJOY THE PODCAST? 

Podcast Transcript

Jon Krohn: 00:05 This is episode number 910 on AI disrupting the journalism industry.

00:19 Welcome back to the SuperDataScience Podcast. I’m your host, Jon Krohn. Back in episode 896 I argued that AI probably won’t be taking your job anytime soon. I followed that up in episode 904 by discussing how some industries are nevertheless being rapidly and thoroughly disrupted by AI. In that episode, 904, I focused on how AI is overhauling the advertising industry in particular. My post announcing that episode on LinkedIn generated a lot of discussion in the comments and garnered over 50,000 impressions within the first few hours of posting, which led me to the idea of having a series of Friday episodes that cover how particular industries, like advertising, are being rapidly and thoroughly overhauled by AI, with lessons for everyone on how we can adapt to this inevitable change and potentially leverage the winds of change to thrive professionally. Today we’re focused on how the journalism industry is being impacted by AI. In future episodes, I’ll cover how AI is disrupting software development, graphic design, education, law, entertainment, and of course, data science itself.

01:24 All right, so starting with journalism. The journalism industry faces its most significant technological transformation since the dawn of the internet as AI fundamentally reshapes news production. Over 81% of journalists now use AI tools in their work, with nearly half using AI daily. While AI promises increased efficiency, it simultaneously raises profound questions about job security, content quality, and the future of human-centered journalism. We’ll cover each of these things in this episode in turn.

01:55 Let’s start with how leading news organizations are integrating AI into their operations. The New York Times launched something called Echo, which we’ve got more info for you in the show notes. It’s an internal summarization tool that provides AI training to all newsroom staff. The Washington Post deployed something called Ask The Post AI, for article summaries, and something else called Haystacker, which analyzed over 700 campaign advertisements in a single investigation, work that would be impossible to do manually.

02:27 Bloomberg, another news outlet, integrated AI summaries into business coverage with its AI research team publishing four papers at EMNLP, Empirical Methods in Natural Language Processing, last year. The BBC, the UK’s great news outlet, said that despite finding that over 50% of AI-generated content contained factual or ethical issues, they nevertheless implemented mandatory AI training and published comprehensive editorial guidelines.

02:56 International innovations include Norway’s iTromso, which I’m probably butchering the name of, building something called D-J-I-N-N, an AI assistant that generated five front page stories in its first week by monitoring municipal documents. And Germany’s EXPRESS.de integrated AI contributions into 11% of articles, achieving 50 to 80% higher click-through rates on that sample.

03:20 Now, how is all of that adoption all over the world impacting job security in journalism? It’s hard to know whether AI is directly responsible for job losses because other factors like the shift to digital or competition from big tech could be playing a part. But last year, nearly 4,000 journalism jobs were lost across the U.K. and the U.S., including an outlet called The Messenger folding, resulting in a loss of 300 jobs, and the giant, Associated Press, cutting 8% of its staff.

03:49 Looking further ahead the U.S. Bureau of Labor Statistics projects a 3% decline in journalism jobs over the next seven or eight years. These employment statistics correlate with widespread adoption of AI in the industry. So, 73% of news organizations use AI for writing tasks. 68% use AI for data analysis overall, and as quickly mentioned at the start of this episode, survey data suggests that four out of every five journalists use AI tools regularly, primarily for tasks like drafting assistance, transcription, fact-checking, and research support.

04:23 If journalists are more productive because of AI, that could explain why news outlets are retaining fewer of them, but it’s not as simple as AI replacing humans. The emergence of hybrid roles, like journalist programmers, suggest a workforce evolution and the AI market in media is projected to reach $100 billion dollars over the next five years. That’s a quadrupling in market size relative to last year, which signals that media organizations view AI as essential infrastructure, fundamentally changing skill requirements across the industry.

04:58 Okay, then, so it’s clear AI is making a big impact on the practice of journalism and will continue to do so, but how is that adoption going? High-profile failures highlight some significant challenges. CNET’s AI-generated financial articles contained embarrassing mathematical errors, including miscalculating simple things like compound interest. I guess, not as simple as simple interest. Sports Illustrated created fictional writers with AI-generated biographies, which violates fundamental journalism principles and resulted in a big backlash.

05:31 Academic research on this also reveals some mixed outcomes. So, academic research showed increased accuracy in routine tasks, for example, but also a reduction in nuance within AI-generated news articles. And I already mentioned earlier how BBC testing found more than half of AI content contained at least some factual or ethical issues. Yet, despite the issues, successful implementations nevertheless demonstrate potential when they are properly managed. German editors, for example, report 94% time savings in transcription. The International Consortium of Investigative Journalists uses AI to analyze millions of leaked documents, identifying patterns impossible for humans to detect. And as I mentioned earlier, the Washington Post’s Haystacker project enabled hitherto impossible political ad analysis, again because of the scale of that investigation.

06:24 So, while AI usage and effectiveness increase, as is often the case, policies are lagging behind the new AI-forward reality. A mere 13% of newsrooms have developed formal AI policies despite the widespread adoption. But policy progress is being made in some governments. The European Union’s AI Act, which becomes effective this month, requires transparency for AI-generated content. California’s AI Transparency Act imposes $5,000 daily penalties for disclosure violations. Despite these early policy strides, however, public trust remains fragile. Pew Research recently found two-thirds of Americans are concerned about inaccurate AI-generated information in the news.

07:06 Okay. So, now you have the lay of the land for today, what’s the future? Well, 87% of newsrooms expect to be fully or somewhat transformed by generative AI in the coming years. That’s basically all of them. The Reuters Institute predicts conversational AI interfaces like ChatGPT and Google’s Gemini will become primary news consumption methods. I think that’s an obvious one. And it’s reasonable to expect that much more change is coming because tons of investment is flowing in. I wasn’t able to find an exact figure, but with AI startups receiving more than half of all global venture capital in early 2025, and with journalism ranking fifth in AI investment focus within Europe at least, it’s a safe bet that at least hundreds of millions of dollars will flow into AI journalism tools in the coming years.

07:55 New business models will continue to emerge around AI-assisted journalism, including revenue sharing with AI platforms like OpenAI and Gemini, as well as AI-powered services available only to paying subscribers. So, that’s how firms will adapt. In terms of what individual journalists can do to prepare themselves for ongoing change, one idea would be to prepare for hybrid roles that combine traditional journalism skills with things like AI literacy and perhaps with software development skills.

08:23 I’m certainly not an expert on journalism, so take my guidance with a grain of salt, but from my research and what I’ve told you in this episode, the evidence appears to reveal an industry in rapid transformation rather than terminal decline. And so, successful newsrooms invest in both AI technology and human development, creating hybrid workflows that combine AI’s processing power with human judgment and ethical reasoning. The future depends on navigating this transformation while maintaining journalism’s essential role in democratic society, balancing innovation with responsibility to ensure AI serves rather than undermines public discourse.

08:58 And if you happen to be a data scientist, AI engineer, or a software developer, and you’ve been interested in getting involved in journalism, it’s never been a better time than now to get involved. Your skills such as sleuthing through data and building interactive visualizations have never been in more demand by the industry. As data volumes continue to grow, demand for your skills should continue to increase for years to come.

09:20 All right. That is my report on how AI is transforming or disrupting the journalism industry. Again, probably not as soon as next week, but over the course of the coming months, I’ll continue to have episodes covering how AI is disrupting different industries, like I mentioned at the onset, things like software development, graphic design, education, law, entertainment, and yes, data science itself.

09:47 All right. That’s it for today’s episode. I’m Jon Krohn and you’ve been listening to the SuperDataScience Podcast. If you enjoyed today’s episode or know someone who might, consider sharing this episode with them, leave a review of the show on your favorite podcasting platform, tag me in a LinkedIn post with your thoughts, and if you aren’t already, obviously subscribe to the show. The most important thing, however, is just keep on listening. Until next time, keep on rocking it out there, and I’m looking forward to enjoying another round of the SuperDataScience Podcast with you very soon.

 

Show All

Share on

Related Podcasts