Exploring collaborative frontiers: AI integration in investigative journalism

In contemporary journalism, Artificial Intelligence (AI) has become an important force, influencing tasks ranging from translations to checking headlines. AI techniques are also increasingly used by investigative journalists, but integration of these techniques comes with challenges. Our study delves into the integration of AI in investigative journalism within the Netherlands, using a multimethod research approach. Through semi-structured interviews with 25 national and regional Dutch news media organisations, including newspapers and broadcasters, we identified a notable challenge: shaping an interdisciplinary news environment where AI experts, legal experts, and journalists collaborate.

While AI is cautiously embraced in all studied newsrooms for investigative journalism, a pervasive lack of technological knowledge, ethical guidelines, and resources hinders its integration. Despite the availability of AI training courses, skills and knowledge acquired are often underutilized or lost over time. We came to understand that journalists, apart from a small group of frontrunners, express uncertainty about tool selection and struggle to comprehensively understand AI-driven research, which they deem crucial for validating their work. A pervasive lack of technological knowledge, ethical guidelines and resources hinders AI integration, the interviews show.

The importance of external expertise is underscored by eight interviewees who explicitly acknowledged the value of hiring or collaborating with individuals possessing in-depth technical knowledge. The lack of legal and ethical expertise is also stressed, with case law around AI and investigative journalism scarcely available. Therefore, journalists are falling back on recognized journalistic codes and ethical frameworks yet struggling with accountability of their AI research findings and the need for more external expertise in this realm. This saves time and gives more confidence in the journalistic AI findings.

"I didn't have the luxury of spending a week figuring out how to do this. An experienced programmer would have written this within an hour."

Communication challenges between journalists and AI experts are highlighted, as journalists find it difficult to articulate their needs to IT professionals, citing language barriers. These struggles are further evident in our Research through Design (RtD) process, where the development of AI prototypes revealed difficulties aligning technical work with journalistic inquiries.

Finally, a lack of resources emerges as a significant barrier, with journalists frequently citing constraints in time for deep dives into AI techniques or hiring AI expertise. A shortage of skilled IT professionals in newsrooms and the limited usability of trained AI models exacerbate these challenges, emphasizing a lack of institutional strategy.

"Technicians look through different glasses and speak a different language."

The importance of multidisciplinary teams in journalism is flagged by a growing number of scholars. Our research underscores the imperative for strategic collaboration between journalists, AI experts, and legal professionals. Despite a noticeable inclination toward experimentation and collaboration, a deficit in expertise, AI literacy, mutual understanding, and resources persists within Dutch AI investigative journalism. Our contribution extends beyond the discourse on responsible AI integration, shedding light on the challenges and opportunities inherent in shaping an interdisciplinary practice within investigative journalism.