Charting the Path for Effective & Ethical AI in Journalism

Nicholas Diakopoulos is a Professor of Communication Studies at the Northwestern School of Communication and Professor of Computer Science (by courtesy) at the McCormick School of Engineering and was a 2023–24 Buffett Faculty Fellow. He is leveraging computational journalism to explore how generative AI will transform newsrooms. 

When ChatGPT first rolled out and quickly scaled up to 100 million users worldwide, it sparked a flurry of speculation about AI's impact on the future of news. How would generative artificial intelligence (AI) models disrupt the journalism industry and change how reporters do their jobs? What would the world look like if AI changed newsrooms without proper ethical guidelines in place? 

As a computer scientist seeking impact in journalism, Northwestern professor and 2023–24 Buffett Faculty Fellow Nicholas Diakopoulos set out to explore these questions and more. He is the principal investigator for the Generative AI + Journalism Initiative, which with co-investigator Professor Jeremy Gilbert is supported by a $1 million grant from the John S. and James L. Knight Foundation to develop new tools and responsible practices for generative AI in news production. He also directs the Northwestern Computational Journalism Lab, which in January 2023 launched the Generative AI in the Newsroom Project dedicated to leveraging AI technology for news media, from data mining and developing story angles to fact-checking and illustrating articles, as well as exploring answers to relevant legal and ethical issues. 

“I take academic ideas and try to get them out into practice and in the hands of industry practitioners,” he said. “There is so much creativity out there, and it underscores how much ground we can cover as scholars when we engage with practice communities.” 

Diakopoulos spoke with us about how generative AI tools are changing journalism and how the Buffett Institute’s faculty community has helped shape this research within Northwestern and beyond.

What inspired the Generative AI in the Newsroom Project?

When ChatGPT was launched in November 2022, I knew it would be big. I wanted to get ideas out quickly and not be in a slow academic peer review cycle. I launched the Generative AI in the Newsroom Project and made a call out to the broader community saying I wanted to help figure out how to use generative AI in responsible ways for news production. I selected about a dozen people’s ideas and committed to working with them as technical consultants and thought partners. As those projects matured, people wrote them up into blog posts on Medium. As an editor, I help folks frame their findings in ways that are valuable to the broader community.

What use cases for generative AI has the project explored?

The project has engaged contributors from around the world, including from various countries in North and South America, Asia and Europe. The range of perspectives offers a rich understanding of how generative AI may be useful in different jurisdictions and legal/policy contexts, and with different levels of resources. By creating a digital space that is truly open to global voices, we’re able to offer a trading zone where readers can also expand their horizon about how people might think about or use generative AI in different parts of the world. 

One contributor wrote about using generative AI to chat with databases so journalists could interact with data more conversationally. Another person built a quiz question generator for news articles to engage audiences. I also have a student who developed a use case to generate different news angles on scientific papers. Thousands of scientific papers come out every month. If we can use tools to help frame what’s newsworthy about them for journalists, we can help science journalists do better journalism. 

Some of the most promising use cases I’ve seen out in the industry have also been the most prosaic. Journalists benefit a lot from having interview recordings automatically transcribed by AI models. Audiences benefit when content can be made more accessible by using generative AI to translate it, either into different languages or between different modalities, such as between an article and a podcast.  

One of the concerns around these AI models is that there may be copyright issues with models trained on copyrighted content. Is that legally or ethically sound? We have since started creating a blueprint to evaluate generative AI models for journalistic use-cases based on quality and ethics. We hope to workshop these ideas with researchers and practitioners at the Computation + Journalism Symposium this fall. 

How do international news outlets compare to U.S. outlets with respect to their adoption of or adaptation to AI?

I’ve found traditional U.S. news organizations to be somewhat behind in re-imagining how generative AI could help with news production. We’ve also seen some of the worst and unethical uses of generative AI in the U.S., with scandals at CNET and Sports Illustrated, to name just a couple. In Europe, many news organizations seem to be investing more in AI, building up multi-functional teams and maturing their understanding of how it can transform the media. I see public media such as BBC, NPO in the Netherlands, Swedish Radio and YLE in Finland investing in understanding and using the technology. These are, of course, broad characterizations, and there are exceptions, but generally I would say Europe is way ahead when it comes to investing in and implementing responsible uses of generative AI in media production.

How did you expand on your research as a Buffett Faculty Fellow? 

One of the things that was exciting to me about the fellowship program was a chance to think more broadly about how my research intersected with different parts of the university and think about more global implications around things like policy.

I've had great conversations with other fellows that showed me the big picture of how something like what’s going on in Silicon Valley creates ripple effects with legal issues or creates new patterns of social behavior among journalists. 

We were also connected with Northwestern’s corporate engagement office through the program, which led to an invited talk at a Silicon Valley research lab. In an important way, the Buffett Faculty Fellowship Program acted as an effective networking vehicle, leading to new and strengthened connections at the university and beyond.